CN103459597A - Marker for predicting prognosis of gastric cancer and method for predicting prognosis of gastric cancer - Google Patents

Abstract

The present invention relates to a marker for predicting gastric cancer prognosis, a composition and a kit for predicting gastric cancer prognosis, which contain an agent for measuring the expression level of the marker, and a method for predicting gastric cancer prognosis using the marker. According to the present invention, gastric cancer prognosis can be accurately predicted, and an appropriate treatment plan can be advantageously established based on the predicted prognosis to significantly reduce death caused by gastric cancer. In particular, according to the present invention, the treatment method for stage III gastric cancer patients can be used for patients who have been predicted to have a negative prognosis among the stage Ib/II gastric cancer patients, whereby the survival rate is greatly improved.

Description

Marker for predicting prognosis of gastric cancer and method for predicting prognosis of gastric cancer

technical field

The present invention relates to a marker for predicting the prognosis of gastric cancer, a composition and a kit for predicting the prognosis of gastric cancer comprising a reagent for measuring the expression level of the marker, and a method for predicting the prognosis of gastric cancer using the marker.

Background technique

In 2005, a total of 65,479 people died of cancer, accounting for 26.7% of all deaths. The cancer that caused the most deaths was lung cancer with 28.4 deaths per 100,000 population (21.1%), followed in order by gastric cancer with 22.6 deaths (16.8%); liver cancer with 22.5 deaths (16.7%); colorectal cancer with 22.5 deaths (16.7%); Cancer, 12.5 patients (9.3%) died. Gastric cancer is known to be the second leading cause of death from cancer worldwide.

Symptoms of stomach cancer show many facets, ranging from asymptomatic to severe pain. Furthermore, stomach cancer symptoms appear to resemble common digestive symptoms without any specific features. Usually, in the early stage of gastric cancer, most cases have no symptoms, and even if there are any symptoms, they are very mild, such as mild indigestion or upper abdominal discomfort, which causes most people to ignore it and thus may increase the death rate of gastric cancer.

The majority of testing methods for gastric cancer have hitherto been physical testing methods. The first choice is gastric X-ray, which includes the double contrast method, compressed X-ray, mucosal mapping, and secondly gastroscopy, which finds very small lesions and allows gastric biopsy at suspicious sites Instead, the diagnostic yield is improved, wherein the very small lesions do not appear when examining the stomach with the naked eye in X-ray examination. However, this method has the disadvantages of hygienic problems and pain experienced by the patient during the examination. Therefore, in recent years, studies on diagnosing gastric cancer by measuring the expression levels of marker genes specifically expressed in the stomach have been conducted, but there are relatively few studies on genetic markers predicting the prognosis of patients with gastric cancer.

The survival rate of patients with gastric cancer depends on the pathological stage at diagnosis. According to Samsung Medical Center, the 5-year survival rate of gastric cancer patients is as follows (KimS et al., Int J Radiat Oncol Biol Phys 2005;63:1279-85).

Phase II: 76.2%, Phase IIIA: 57.6%,

Stage IIIB: 39.6%, Stage IV 26.3%

The results showed that early detection of gastric cancer can significantly improve the survival rate. However, since gastric cancers that have been diagnosed at the same stage show differences in prognosis depending on patients, accurate prediction of gastric cancer prognosis and early detection of gastric cancer are the most important factors for effective treatment of gastric cancer.

On the other hand, in order to diagnose gastric cancer, the doctor starts the necessary examination of the patient, which is considered to be the most appropriate treatment plan. There are methods for treating cancer such as surgery, endoscopic therapy, chemotherapy and radiation therapy. Treatment is generally determined by considering stomach cancer therapy, the size, location and extent of the stomach cancer, the patient's general health, and many other factors.

In the case of stage IB/II gastric cancer treated only with surgery, approximately 30% of patients are known to relapse within 5 years. In this case, since it is impossible to predict in which patients gastric cancer will recur, different doctors use different treatments. Therefore, if the prognosis of gastric cancer patients can be accurately predicted, an appropriate treatment such as surgery or chemotherapy can be determined based on the prognosis, which may greatly contribute to the survival of gastric cancer patients, and thus a technology that can accurately predict the prognosis of gastric cancer patients is required.

Conventionally, anatomical observation (the degree of cancer cell invasion and the number of metastatic lymph nodes) has been used in order to predict the prognosis of gastric cancer patients, but this method has limitations in possible intervention of physician's subjective judgment and accurate prediction of prognosis.

Contents of the invention

invention disclosure

Technical Problem In this context, as a result of research that the survival rate of gastric cancer patients can be improved by accurately predicting the prognosis of gastric cancer and determining an appropriate direction of treatment based on the predicted prognosis, the present inventors determined that it is possible to identify a marker for predicting the prognosis of gastric cancer and The expression levels of the markers are measured to accurately predict the prognosis of gastric cancer, so as to complete the present invention.

Technical solutions

The object of the present invention is to provide a marker for predicting the prognosis of gastric cancer, said marker comprising one or more genes selected from the following: C20orf103, COL10A1, MATN3, FMO2, FOXS1, COL8A1, THBS4, CDC25B, CDK1, CLIP4, LTB4R2, NOX4, TFDP1, ADRA2C, CSK, FZD9, GALR1, GRM6, INSR, LPHN1, LYN, MRGPRX3, ALAS1, CASP8, CLYBL, CST2, HSPC159, MADCAM1, MAF, REG3A, RNF152, UCHL1, ZBED5, GPNMB, HIST1H2AJ, RPL9, DPP6, ARL10, ISLR2, GPBAR1, CPS1, BCL11B, and PCDHGA8 genes.

Another object of the present invention is to provide a composition for predicting the prognosis of gastric cancer, the composition comprising reagents for measuring mRNA or protein expression levels of markers for predicting the prognosis of gastric cancer.

Another object of the present invention is to provide a kit for predicting the prognosis of gastric cancer, the kit comprising reagents for measuring mRNA or protein expression levels of markers for predicting the prognosis of gastric cancer.

Another object of the present invention is to provide a method for predicting the prognosis of gastric cancer, the method comprising a) obtaining the expression level or expression pattern of mRNA or protein markers used to predict the prognosis of gastric cancer in samples collected from gastric cancer patients; and b) comparing the expression level or expression pattern obtained from step a) with the expression level or expression pattern of mRNA or protein of the corresponding gene in gastric cancer patients with known prognosis.

Another object of the present invention is to provide a method for predicting the prognosis of gastric cancer, the method comprising a) measuring the expression level or expression pattern of mRNA or protein for predicting the prognosis of gastric cancer in a sample collected from a gastric cancer patient, obtaining a quantitative expression value; b) applying the expression value obtained in step a) to a prognosis prediction model to obtain a gastric cancer prognosis score; and c) comparing the gastric cancer prognosis score obtained in step b) with a reference value to determine the patient's prognosis.

Beneficial effect

According to the present invention, the prognosis of gastric cancer can be predicted rapidly and accurately, and appropriate treatment can be determined based on the predicted prognosis, which has the advantage of contributing to a significant reduction in death caused by gastric cancer. In particular, according to the present invention, the survival rate can be greatly improved by using the targeted therapy developed for stage III gastric cancer, because among patients with stage Ib/II gastric cancer, those who have been predicted to have a negative prognosis show a similar prognosis as that of patients with stage III gastric cancer And resistant to existing standard chemotherapy.

Brief description of the drawings

Figure 1 is a graph showing the relationship between various risks based on quantile normalization and self normalization using a reference gene.

Figure 2 represents Kaplan-Meier curves according to the expression levels of C20orf103, COL10A1 genes.

Fig. 3 represents Kaplan-Meier curves according to the expression levels of MATN3, FMO2 genes.

Fig. 4 represents Kaplan-Meier curves according to the expression levels of FOXS1, COL8A1 genes.

Fig. 5 represents Kaplan-Meier curves according to the expression levels of THBS4, ALAS1 genes.

Fig. 6 represents Kaplan-Meier curves according to the expression levels of CASP8, CLYBL genes.

Fig. 7 represents Kaplan-Meier curves according to the expression levels of CST2, HSPC159 genes.

Fig. 8 represents Kaplan-Meier curves according to the expression levels of MADCAM1, MAF genes.

Fig. 9 represents Kaplan-Meier curves according to the expression levels of REG3A, RNF152 genes.

Fig. 10 represents Kaplan-Meier curves according to the expression levels of UCHL1, ZBED5 genes.

Fig. 11 represents Kaplan-Meier curves according to the expression levels of GPNMB, HIST1H2AJ genes.

Fig. 12 represents Kaplan-Meier curves according to the expression levels of RPL9, DPP6 genes.

Fig. 13 represents Kaplan-Meier curves according to the expression levels of ARL10, ISLR2 genes.

Fig. 14 represents Kaplan-Meier curves according to the expression levels of GPBAR1, CPS1 genes.

Figure 15 represents Kaplan-Meier curves according to the expression levels of BCL11B, PCDHGA8 genes.

The p-values in FIGS. 2 to 15 are the result values of the log-rank test by dividing the expression levels of genes by means of high expression or low expression and gene expression levels.

16 is a Kaplan-Meier curve showing disease-free survival for positive prognosis groups (low risk) or negative prognosis groups (high risk), divided according to the prognosis prediction model using the genes listed in Table 5. FIG.

17 is a Kaplan-Meier curve showing the disease-free survival rate of Ib/II stage gastric cancer patients divided into positive prognosis group or negative prognosis group using the prognostic prediction model of the genes listed in Table 5 .

18 is a Kaplan-Meier curve showing disease-free survival for positive prognosis groups (low risk) or negative prognosis groups (high risk), divided according to the prognosis prediction model using the genes listed in Table 7. FIG. HR in Figure 18 is the cumulative hazard function ratio and p-values were calculated using 100 permutations.

Fig. 19 is the Kaplan-Meier curve of the patient group after dividing the patients (high vs low), wherein according to the prognostic prediction model using the genes listed in Table 7 and according to the pathological stage (IB+II vs III+IV) patient classification. P-values were calculated by two-sided log-rank test.

Fig. 20 represents Kaplan-Meier curves according to the expression levels of CDC25B, CDK1 genes.

Fig. 21 represents Kaplan-Meier curves according to the expression levels of CLIP4, LTB4R2 genes.

Fig. 22 represents Kaplan-Meier curves according to the expression levels of NOX4, TFDP1 genes.

Fig. 23 represents Kaplan-Meier curves according to the expression levels of ADRA2C, CSK genes.

Fig. 24 represents Kaplan-Meier curves according to the expression levels of FZD9, GALR1 genes.

Fig. 25 represents Kaplan-Meier curves according to the expression levels of GRM6, INSR genes.

Fig. 26 represents Kaplan-Meier curves according to the expression levels of LPHN1, LYN genes.

Fig. 27 represents a Kaplan-Meier curve according to the expression level of the MRGPRX3 gene.

The p-values in FIGS. 20 to 27 are the result values of the log-rank test by dividing the expression levels of genes by means of high expression or low expression and gene expression levels.

Figure 28 represents the cutoff analysis of the GCPS for the genes listed in Table 10. The best discrimination is the case where patients are divided into 75% high-risk group and 25% low-risk group.

Figure 29 represents the disease-free survival of stage II gastric cancer patients in the discovery set based on the optimized cut-off values of GCPS for the genes listed in Table 10.

Figure 30 represents the distribution of GCPS for the genes listed in Table 10 in the discovery set versus the validation set, and shows that the distribution of GCPS in the discovery set coincides with the distribution of GCPS in the discovery set. This represents the analytical robustness of this assay.

Figure 31 represents disease-free survival for the validation cohort according to the predefined algorithm GCPS and cut-off values (red = high risk).

32 represents the disease-free survival rate of stage II gastric cancer patients who underwent surgery and radiation therapy based on the GCPS of the genes listed in Table 11. Blue color represents high risk as defined by GCPS.

Figure 33 represents the disease-free survival rate of stage II gastric cancer patients who underwent surgery based only on the GCPS of the genes listed in Table 11. Blue color represents high risk as defined by GCPS.

Detailed ways

Preferred Invention Mode

As an aspect of achieving the goal, the present invention provides markers for predicting the prognosis of gastric cancer, which include one or more genes selected from the following genes: C20orf103, COL10A1, MATN3, FMO2, FOXS1, COL8A1, THBS4, CDC25B, CDK1, CLIP4, LTB4R2, NOX4, TFDP1, ADRA2C, CSK, FZD9, GALR1, GRM6, INSR, LPHN1, LYN, MRGPRX3, ALAS1, CASP8, CLYBL, CST2, HSPC159, MADCAM1, MAF, REG3A, RNF152, UCHL1, ZBED5, GPNMB, HIST1H2AJ, RPL9, DPP6, ARL10, ISLR2, GPBAR1, CPS1, BCL11B, and PCDHGA8 genes.

As another aspect, the present invention provides a composition for predicting the prognosis of gastric cancer, the composition comprising a reagent for measuring the expression level of mRNA or protein of a marker for predicting the prognosis of gastric cancer.

Although in the same pathological stage, the clinical prognosis of each gastric cancer is different, and appropriate treatment methods must be used according to this prognosis in order to increase the survival rate of gastric cancer patients. Thus, the present invention provides a composition for predicting the prognosis of gastric cancer, the composition comprising a marker for predicting the prognosis of gastric cancer and a reagent for measuring the expression level thereof, in order to accurately predict the prognosis of a patient diagnosed with gastric cancer And determine the appropriate treatment direction based on the predicted prognosis to increase the survival rate of gastric cancer patients.

As used herein, the term "marker" refers to a molecule that is quantitatively or qualitatively related to the presence of a biological phenomenon, and the marker of the present invention refers to a gene that serves as a basis for predicting the presence of good or poor prognosis in gastric cancer patients.

The markers of the present invention have significantly low p-values and high reliability for predicting the prognosis of gastric cancer, and in particular, the markers listed in Tables 5, 7, 10 and 11 can divide patients into groups according to the expression levels of the markers. Divided into a positive prognosis group or a negative prognosis group, and the prognosis of gastric cancer patients can be accurately predicted by measuring the expression levels of the markers, because according to the Kaplan-Meier curve showing the survival rates of these groups, the survival rate of the positive prognosis group is higher than that of the negative prognosis group group survival.

As used herein, the term "prognosis" refers to expectations regarding medical development (e.g., likelihood of long-term survival, disease-free survival, etc.), including a positive prognosis or a negative prognosis including disease progression such as recurrence, tumor growth, metastasis and resistance mortality (mortality), and positive outcomes include disease remission, such as disease-free status, and disease improvement, such as tumor regression or stabilization (stabilization).

As used herein, the term "prediction" refers to guesses about medical developments, and for the purposes of the present invention, guesses about disease progression (disease progression, improvement, gastric cancer recurrence, tumor growth, drug resistance) in patients diagnosed with gastric cancer.

In an example of the present invention, the prognosis of gastric cancer patients is predicted by dividing patients diagnosed with gastric cancer into a positive prognosis group or a negative prognosis group, and additionally, by dividing patients diagnosed with gastric cancer pathological stages according to the prognosis. Prognosis of gastric cancer patients (Examples 7 to 9).

The markers for predicting the prognosis of gastric cancer can preferably be a combination of the following genes: C20orf103, COL10A1, MATN3, FMO2, FOXS1, COL8A1 and THBS4 genes; a combination of the following genes: ALAS1, C20orf103, CASP8, CLYBL, COL10A1, CST2, FMO2, FOXS1, HSPC159, MADCAM1, MAF, REG3A, RNF152, THBS4, UCHL1, ZBED5, GPNMB, HIST1H2AJ, RPL9, DPP6, ARL10, ISLR2, GPBAR1, CPS1, BCL11B, and PCDHGA8 genes; combination of the following genes: C20orf103, CDC25B, CDK1, CLIP4, LTB4R2, MATN3, NOX4 and TFDP1 genes; or a combination of the following genes: ADRA2C, C20orf103, CLIP4, CSK, FZD9, GALR1, GRM6, INSR, LPHN1, LYN, MATN3, MRGPRX3 and NOX4 genes; and more preferably the following A combination of genes: C20orf103, CDC25B, CDK1, CLIP4, LTB4R2, MATN3, NOX4, and TFDP1 genes, or a combination of the following genes: ADRA2C, C20orf103, CLIP4, CSK, FZD9, GALR1, GRM6, INSR, LPHN1, LYN, MATN3, MRGPRX3 and the NOX4 gene.

The present inventors determined that the above genes can accurately predict the prognosis of gastric cancer by the following method. The present inventors extracted RNA from formalin-fixed paraffin-embedded gastric cancer tumor tissues, measured gene expression levels using the extracted RNA and a Whole-Genome DASL assay kit, and then used the Gene expression levels were treated as a continuous variable with a Cox proportional hazard model for standard statistical analysis. As a result, 369 genes were identified for predicting the prognosis of gastric cancer by Univariate analysis (Table 2) with a large correlation with disease-free survival, and for predicting the prognosis of pathological stage Ib/II gastric cancer genes (Table 3). Subsequently, by applying the superPC algorithm to the expression levels of the identified genes, a prognostic prediction model containing the genes in Table 5 was generated, and according to this prediction model, gastric cancer patients were divided into a positive prognosis group or a negative prognosis group. By showing that the survival rate of the positive prognosis group is higher than that of the negative prognosis group (Example 7 and Figure 16, Figure 17), the Kaplan-Meier curve verifies the effectiveness and reliability of the prognosis prediction model using the markers of the present invention for the results of dividing groups . In addition, the result of dividing gastric cancer patients into a positive prognosis group or a negative prognosis group by applying the gradient lasso algorithm (gradient lasso algorithm) to the expression levels of the identified genes to generate a prognosis prediction model containing the genes in Table 7 determined this Classification coincides with clinical outcomes (Example 8 and Figures 18, 19).

As used herein, the term "a reagent for measuring the expression level of a marker" refers to a molecule that can be used to determine the expression level of marker genes or proteins encoded by these genes, and may preferably be an antibody specific for said marker, Primers or probes.

As used herein, the term "antibody" is an art-recognized term referring to a specific protein molecule directed against an antigenic site. For the purposes of the present invention, an antibody refers to an antibody that specifically binds to a marker of the present invention and can be prepared by conventional methods from proteins encoded by marker genes obtained by cloning each gene into an expression vector in a conventional manner . Among them, partial peptides that can be produced from the protein are included.

As used herein, the term "primer" refers to a short nucleic acid sequence, as a nucleic acid sequence with a short free 3' terminal hydroxyl group (free 3' hydroxyl), which can form a base pair with a complementary template (template) and serve as The starting point for copying templates. In the present invention, the prognosis of gastric cancer can be predicted by whether or not a desired product is produced by performing PCR amplification using sense and antisense primers of the marker polynucleotide of the present invention. PCR conditions and lengths of sense and antisense primers can be modified based on what is known in the art.

As used herein, the term "probe" refers to nucleic acid fragments such as RNA or DNA as short as a few to several hundred bases, which can establish specific binding with mRNA and can maintain labeling (Labeling) Determine the presence of a specific mRNA. Probes can be prepared in the form of oligonucleotide probes, single-stranded DNA (single stranded DNA) probes, double-stranded DNA (double stranded DNA) probes, and RNA probes. In the present invention, the prognosis of gastric cancer can be predicted by hybridization using the labeled polynucleotide of the present invention and a complementary probe. Proper selection of probes and hybridization conditions can be modified based on what is known in the art.

The primers or probes of the present invention can be chemically synthesized using phosphoramidite solid phase support method or other well-known methods. The nucleic acid sequences may also be modified using a number of means known in the art. Non-limiting examples of such modifications are methylation, capping, substitution with one or more analogs of natural nucleotides and modifications between nucleotides, e.g., modification of uncharged linkers (e.g., methyl phosphate, phosphotriester, phosphorimide, carbamate, etc.), or modify charged linkers (e.g., phosphorothioate, phosphorodithioate, etc.).

In the present invention, the expression level of the marker for predicting the prognosis of gastric cancer can be determined by determining the expression level of the mRNA of the marker gene or the protein encoded by the gene.

As used herein, the term "measuring the expression level of mRNA" refers to the process of determining the presence and expression level of mRNA of a marker gene in a biological sample in order to predict the prognosis of gastric cancer and can be achieved by measuring the amount of mRNA. Analytical methods used for this purpose are but not limited to RT-PCR, competitive RT-PCR (competitive RT-PCR), real-time RT-PCR (Real-time RT PCR), RNase protection assay (RPA; RNase protection assay) , northern blotting, DNA microarray chips, etc.

As used herein, the term "measuring the expression level of a protein" refers to the process of determining the presence and expression level of a protein expressed in a marker gene in a biological sample in order to predict the prognosis of gastric cancer, and can be obtained by using a protein that specifically binds to the protein expressed in the above-mentioned gene. Antibodies to determine the amount of protein. Analytical methods used for this purpose are but not limited to western blotting, ELISA (enzyme-linked immunosorbent assay), radioimmunoassay, radioimmunodiffusion, Ouchterlony Immunodiffusion, Rocket electrophoresis, tissue immunostaining, immunoprecipitation assay, complete fixation assay, FACS, protein chip, etc.

As another aspect, the present invention provides a kit for predicting the prognosis of gastric cancer, the kit comprising a reagent for measuring the expression level of mRNA or protein of a marker used for predicting the prognosis of gastric cancer.

The kit of the present invention can be used to identify the expression levels of markers for predicting the prognosis of gastric cancer so as to predict the prognosis of gastric cancer.

The kit of the present invention may be a RT-PCR kit, a real-time RT-PCR kit, a real-time QRT-PCR kit, a microarray chip kit or a protein chip kit.

The kit of the present invention may not only contain primers and probes for measuring the expression level of markers for predicting the prognosis of gastric cancer, or antibodies that specifically recognize the markers, but also include one or more types of other components suitable for the analysis method compositions, solutions or devices.

According to an example of the present invention, the kit for measuring the expression level of mRNA of a marker gene may be a kit containing essential elements required for performing RT-PCR. In addition to each pair of primers specific for a marker gene, this RT-PCR kit can also contain test tubes or other suitable containers, reaction buffer solution, deoxynucleotides (dNTPs), Taq-polymerase and reverse transcriptase, DNA Enzymes, RNase inhibitors and DEPC water (DEPC-water) and sterile water.

According to another example of the present invention, the kit for measuring the expression level of the protein encoded by the marker gene may comprise a substrate, a suitable buffer solution, a second antibody labeled with a chromogenic enzyme or a fluorescent substance, and a chromogenic substrate.

According to another example of the present invention, the kit in the present invention may be a kit for detecting markers for predicting the prognosis of gastric cancer, which contains the necessary elements required for performing DNA microarray chips. The DNA microarray chip kit may comprise a substrate to which a gene as a probe or a cDNA corresponding to a fragment thereof is ligated, and the substrate may include a quantitative control gene or a cDNA corresponding to a fragment thereof.

As another aspect, the present invention provides a method for predicting the prognosis of gastric cancer, the method comprising a) obtaining the expression level or expression pattern of mRNA or protein markers used to predict the prognosis of gastric cancer in samples collected from gastric cancer patients; and b) comparing the expression level or expression pattern obtained in step a) with the expression level or expression pattern of mRNA or protein of the corresponding gene in gastric cancer patients with known prognosis.

As used herein, the term "sample collected from a gastric cancer patient" may be, but not limited to, tissue, cells, whole blood, serum, plasma, and preferably gastric tumor tissue derived from the stomach of a gastric cancer patient.

As used herein, the term "gastric cancer patient with known prognosis" refers to a patient whose disease progression has been revealed among patients diagnosed with gastric cancer, for example, a patient confirmed to have a negative prognosis due to recurrence within A patient whose prognosis is confirmed to be completely cured, and the prognosis of the patient whose prognosis is to be found can be accurately predicted by obtaining and comparing the expression level or expression pattern from the sample collected from the above-mentioned patient and the patient whose prognosis is to be found.

According to an example of the present invention, prognosis can be predicted by measuring the expression levels or expression patterns of marker genes from many gastric cancer patients, using the patient's prognosis to build a database of measured values, and comparing the expression of the patient whose prognosis is to be found Levels or expression patterns are entered into the database. In such cases, known algorithms or statistical analysis programs can be used to compare expression levels or patterns. In addition, the database can be further subdivided into pathology stage, treatment received, etc.

According to an example of the present invention, the gastric cancer patient in steps a) and b) is a patient receiving the same treatment, and said treatment may be radiotherapy, chemotherapy, chemoradiotherapy, adjuvant chemotherapy, gastrectomy, gastrectomy Gastrectomy without radiation therapy after resection with chemotherapy or chemoradiation and adjuvant chemotherapy or surgery.

According to an example of the present invention, the gastric cancer may be stage Ib or stage II gastric cancer.

In the present invention, the expression level of a marker gene can be measured at the level of mRNA or protein, and mRNA or protein can be isolated from a biological sample using a publicly known method.

Analytical methods for measuring levels of mRNA or protein are as described above.

By means of the above analysis method, the expression level of gastric cancer gene markers measured from samples of gastric cancer patients whose prognosis is known can be compared with the expression level of gastric cancer gene markers measured from samples of patients whose prognosis is to be found, and can be determined by determining The increase or decrease of the expression level can predict the prognosis of gastric cancer. In other words, if a sample of a patient whose prognosis is to be found shows a similar expression level or expression pattern to a sample of a gastric cancer patient with a positive prognosis by comparing the results of expression levels, it can be determined to have a positive prognosis, and conversely, if it shows A similar expression level or expression pattern to samples from gastric cancer patients with a negative prognosis can be determined to have a negative prognosis.

According to an example of the present invention, the prognosis can be predicted by comparing and normalizing the expression level of the marker gene with the expression level of one or more genes selected from the genes listed in Table 4, and then using the normalization Expression level of normalization.

As another aspect, the present invention provides a method for predicting the prognosis of gastric cancer, the method comprising a) measuring the expression level of mRNA or protein of a marker used to predict the prognosis of gastric cancer in a sample collected from a gastric cancer patient to obtain a quantitative an expression value; b) applying the expression value obtained in step a) to a prognosis prediction model to obtain a gastric cancer prognosis score; and c) comparing the gastric cancer prognosis score obtained in step b) with a reference value to determine the prognosis of the patient.

Step a) is a step for quantitatively measuring the expression level of the marker gene. Quantitative expression values for marker genes can be obtained using known software, kits and systems to quantify the expression levels measured by the assay methods described above for measuring mRNA or protein levels. According to an example of the present invention, measuring the expression level of a marker gene can be performed using nCounter Assay Kit (NanoString Technologies). In this case, the expression level of the marker gene can be normalized by comparison with the expression level of the reference gene. According to an example of the present invention, the measured marker gene expression level can be normalized by comparing with the expression level of one or more reference genes selected from the reference genes listed in Table 4.

According to an example of the present invention, in step a), C20orf103, CDC25B, CDK1, CLIP4, LTB4R2, MATN3, NOX4 and TFDP1 genes, or ADRA2C, C20orf103, CLIP4, CSK, FZD9, GALR1, GRM6, INSR, LPHN1, Expression levels of mRNA or protein of LYN, MATN3, MRGPRX3 and NOX4 genes.

Step b) is a step of applying the expression value obtained in step a) to a prognosis prediction model to obtain a gastric cancer prognosis score.

According to the example of the present invention, this prognosis prediction model can be expressed as:

[S=β ₁ x ₁ +...+β _n x _n ]

Among them, x _n is the quantitative expression value of the nth gene,

β _n is the Cox Regression estimate for the nth gene, and

S stands for gastric cancer prognosis score.

Step c) is a step of comparing the gastric cancer prognosis score obtained in step b) with a reference value to determine the prognosis of the patient.

The reference value can be determined as a value within the range of the cut-off value of the third quartile (third quartile) to the cut-off value of the fourth quartile in the multiple gastric cancer prognostic score distribution, wherein by inputting from multiple The expression values of marker genes of gastric cancer patients are used to obtain the multiple gastric cancer prognosis score distributions. In addition, the reference value can be determined as a value within the range from the cut-off value of the second quartile (Second quartile) to the cut-off value of the third quartile in the distribution of multiple gastric cancer prognostic scores, wherein by inputting The expression values of marker genes from a plurality of gastric cancer patients are used to obtain the distribution of the multiple gastric cancer prognosis scores. Preferably, the reference value can be determined as a value within the range of the cut-off value of the third quartile to the cut-off value of the fourth quartile in the distribution of multiple gastric cancer prognostic scores, wherein by inputting The expression values of marker genes of gastric cancer patients are used to obtain the multiple gastric cancer prognosis score distributions.

The cutoff values of quartiles can be defined as values corresponding to 1/4, 2/4, 3/4, and 4/4 points when a plurality of gastric cancer patients are distributed according to the scale of the gastric cancer prognostic score. In this case, the cut-off value of the fourth quartile may be the largest score among the gastric cancer prognostic scores obtained from the patient.

According to an example of the present invention, it can be determined that a case with a gastric cancer prognosis score obtained in step b) that is equal to or greater than the reference value has a negative prognosis.

According to an example of the present invention, the critical value can be 0.2205 or -0.4478, and it can be determined that the cases with the gastric cancer prognostic score obtained in step b) are equal to or greater than the critical value have a negative prognosis. Preferably, the cut-off value may be 0.2205 if the expression levels of C20orf103, CDC25B, CDK1, CLIP4, LTB4R2, MATN3, NOX4 and TFDP1 genes are measured in step a), and if ADRA2C, C20orf103, CLIP4 are measured in step a) , CSK, FZD9, GALR1, GRM6, INSR, LPHN1, LYN, MATN3, MRGPRX3 and NOX4 gene expression levels, the critical value can be -0.4478.

In an example of the present invention, the prognosis prediction model comprising the genes in Table 10 and Table 11 is generated by applying the gradient lasso algorithm, and gastric cancer patients are divided into positive prognosis group or negative prognosis group by comparing the gastric cancer prognosis value, wherein by Enter the expression value and reference value into the above formula to obtain the gastric cancer prognosis value. The Kaplan-Meier curve results for the divided groups validated the use of Validity and reliability of the prognostic prediction model marked by the present invention. In addition, as a result of dividing the patients according to the gastric cancer prognosis value obtained by measuring the expression level of marker genes with patients who underwent gastrectomy alone as subjects, it was determined that by confirming that the survival rate of the negative prognosis group (high risk) was significantly lower, The markers of the invention can also be used to predict prognosis in patients undergoing gastrectomy alone (Example 9 and Figure 33).

Therefore, the prognosis of gastric cancer can be accurately predicted according to the present invention, and the benefit of an appropriate treatment regimen consistent with the predicted prognosis can be obtained. For example, standard therapy or less invasive treatment options may be determined for patients judged to have a positive prognosis, and treatment options for early gastric cancer patients or very invasive treatment options may be determined for patients judged to have a negative prognosis. treatment) or experimental therapy. Specifically, for patients diagnosed with stage Ib or II gastric cancer, an appropriate treatment method can be selected according to the prognosis predicted by the present invention, because these patients may show different prognosis. For example, treatments such as surgery or anticancer drugs for patients with stage III gastric cancer can be used in patients diagnosed with stage Ib or II gastric cancer who are predicted to have a negative prognosis.

way of invention

Hereinafter, the present invention is described in more detail by providing examples. However, these examples are intended to be illustrative only, and not to limit the claimed invention in any way.

Example 1: Selection of Gastric Cancer Patients

This study was conducted at Samsung Medical Center and Samsung Cancer Research Institute in accordance with the Declaration of Helsinki. This study was approved by the Steering Committee of Samsung Medical Center. Between 1994 and December 2005, a cohort of 1152 patients was selected from 1557 patients eligible for gastrectomy after adjuvant chemotherapy with 5-FU/LV (INT-0116 regimen) according to the following criteria:

1) Histological diagnosis of adenoma, tumor resection, no residual tumor,

2) D2 lymph node dissection (D2 lymph node dissection),

3) Men and women over the age of 18,

4) Pathological stage Ib (T2bN0, T1N1 or not T2aN0) to IV according to AJCC (American Joint Committee on Cancer) 6th edition,

5) Patients who kept complete surgical records and treatment records, and received 5-fluorouracil/leucovorin (5-fluorouracil/leucovorin) adjuvant chemotherapy (INT-0116 program) at least twice according to the following methods. That is, patients who received chemoradiation (total radiation of 4500 cGy, 180 cGy per day, 1 week/5 days, for 5 weeks) followed by administration of 5-fluorouracil (400 mg/m ² /day) and leucovorin ( 20mg/m ² /day) for 5 days (once), and an additional administration of 5-fluorouracil (400mg/m ² /day) and leucovorin (20mg/m ² /day).

405 patients in a group of 1557 patients were excluded from this analysis due to the following reasons:

1) Patients who received less than 2 times of 5-FU/LV adjuvant chemotherapy (N=144),

2) Patients with positive resection margin under the microscope (N=73),

3) Patients with double primary cancer (N=53),

4) Patients with recurrence of gastric cancer in the remnant stomach after subtotal gastrectomy (N=5),

5) Patients without complete medical records (N=11),

6) Patients using regimens other than INT-0116 regimen (N=65)

7) Others (N=54).

This study was carried out with 432 patients, which were finally randomly screened after secondary screening of 1152 patients from 1557 primary screened patients, and the medical characteristics of the patients are shown in Table 1. The classification of the 432 patients according to the pathological stages of gastric cancer showed the following composition: 68 in stage Ib, 167 in stage II, 111 in stage IIIA, 19 in stage IIIB, and 67 in stage IV (Table 1).

Table 1

Example 2: RNA extraction from gastric tumors

FFPE RNA分离试剂盒（Omega Bio-Tek,Norcross,GA,美国）提取完整RNA。使用NanoDrop8000分光光度计（Thermo Scientific）测定提取的RNA的浓度，并且将其在使用之前贮存在-80°C的低温。在实验中，作为不适宜的样品，浓度小于40ng/μl并且A260/A280比率小于1.5或A260/230比率小于1.0的RNA样品不用分析中。RNA was extracted from gastric tumors of gastric cancer patients finally screened in Example 1. For this purpose, the primary tumor paraffin block consisting of the largest tumor was selected. RNA was extracted from 2 to 4 sections of 4 μm thickness in formalin-fixed, paraffin-embedded tissue and pipetted by microdissection before transfer to extraction tubes. Unless the tumor element. Subsequently, according to the manufacturer's instructions, using the High Pure RNA Paraffin kit (Roche Diagnostic, Mannheim, Germany) or EZNA

FFPE RNA Isolation Kit (Omega Bio-Tek, Norcross, GA, USA) was used to extract intact RNA. The concentration of extracted RNA was determined using a NanoDrop 8000 spectrophotometer (Thermo Scientific) and stored at -80°C until use. In the experiment, RNA samples with a concentration of less than 40 ng/μl and an A260/A280 ratio of less than 1.5 or an A260/230 ratio of less than 1.0 were not analyzed as unsuitable samples.

Example 3: Whole genome expression profiling

（Illumina Whole-Genome DASL

）（cDNA介导的复性、选择、延长和连接，Illumina，美国）测定法。首先，通过以下方式制备PCR模板：使用生物素化的寡-dT（biotinylatedoligodT）引物和随机引物（random primers）将完整RNA逆转录成cDNA，使生物素化的cDNA与一对查询寡聚物（query oligos）复性，延伸查询寡聚物之间的空位并且随后连接。随后，使用一对通用PCR引物（universal PCR primers）扩增的PCR产物与人Ref-8表达珠芯片（humanRef-8Expression BeadChip）（>24,000种注释的转录物）杂交。在杂交后，使用iScan（Illumina，美国）扫描人Ref-8珠芯片。Illumina genome-wide DASL was performed with 200 ng of RNA extracted in Example 2 according to the manufacturer's instructions

(Illumina Whole-Genome DASL

) (cDNA-mediated renaturation, selection, elongation and ligation, Illumina, USA) assay. First, PCR templates were prepared by reverse-transcribing intact RNA into cDNA using biotinylated oligo-dT (biotinylated oligodT) primers and random primers, and combining biotinylated cDNA with a pair of query oligos ( query oligos), the gaps between the query oligos are extended and subsequently ligated. Subsequently, PCR products amplified using a pair of universal PCR primers were hybridized to the humanRef-8 Expression BeadChip (>24,000 annotated transcripts). After hybridization, the human Ref-8 bead chip was scanned using iScan (Illumina, USA).

Example 4: Quality control of Whole-Genome DASLassay

Probes called "absent" among 24,526 probes of the human Ref-8 expression bead chip used in Example 3 were filtered and removed. The 17,418 probes remaining after filtering were used in later analysis. Probe intensities were corrected by logarithm with base2) and normalized using a quantile normalization algorithm. As a result, statistical analysis was performed using 17,418 probes and 432 samples.

Example 5: Identification of gastric cancer predictive genes

To identify genes whose expression levels correlate with clinical outcomes such as disease free survival (DFS), standard statistical analysis was performed using a Cox proportional hazard model, with treatment as a continuous variable (continuous variables) gene expression levels. As a result, 369 probes having a significant correlation with the disease-free survival rate among 17,418 probes were identified by Univariate analysis, and the results are shown in Table 2 (p<0.001).

In addition, since it is important to predict the prognosis of patients with stage Ib/II (stage Ib/II), using samples collected from patients with stage Ib/II among the samples as objects, in the same manner as above, stage Ib/II-specific Gastric cancer prognosis genes, and the results are shown in Table 3. The p value (p value) in Table 3 represents the degree of influence of gene expression level on clinical prognosis, among which the lower p value affects the prognosis more significantly, and the hazard ratio represents the degree of influence on the recurrence rate of gastric cancer, and the increase or decrease of the number has significant meaning.

According to Tables 2 and 3, the presence of numerous stage Ib/II-specific prognostic genes was identified, but the prognostic genes identified using the entire cohort of patients coincided with those identified using stage Ib/II patients .

Table 2

table 3

Primer number symbol Login ID P value risk ILMN_1736078 THBS4 NM_003248.3 1.38572E-06 1.682498015 ILMN_1713561 C20orf103 NM_012261.2 2.55588E-06 1.478694678 ILMN_1776490 C17orf53 NM_024032.2 6.03826E-06 0.34241099 ILMN_1755318 HIST1H2AJ NM_021066.2 1.2522E-05 0.470375569 ILMN_1769168 ARL10 NM_173664.4 1.4066E-05 1.561604462 ILMN_2180606 NAT13 NM_025146.1 1.73444E-05 0.33114386 ILMN_1726815 HIST1H3G NM_003534.2 2.23113E-05 0.599169194 ILMN_1663786 EPB41 NM_203342.1 5.97621E-05 0.488778032 ILMN_1789955 PNRC1 NM_006813.1 7.41079E-05 1.448627279 ILMN_1762003 SEC62 NM_003262.3 8.06626E-05 0.374133795 ILMN_1757060 CAMK2D NM_172115.1 8.99236E-05 0.276235186 ILMN_1721127 HIST1H3D NM_003530.3 0.000100991 0.360536141 ILMN_2390544 DKFZP564J102 NM_015398.2 0.000123879 1.538277475 ILMN_2249018 LOC389816 NM_001013653.1 0.000139453 0.427195449 ILMN_1694877 CASP6 NM_001226.3 0.000139804 0.325712607 ILMN_2103685 DEPDC1B NM_018369.1 0.00014085 0.366505298 ILMN_1694472 GCK NM_033508.1 0.000145382 1.428117308 ILMN_1769207 KCTD7 NM_153033.1 0.000190952 1.585238708 ILMN_1738116 TMEM119 NM_181724.1 0.000198141 1.489761712 ILMN1725314 GBP3 NM_018284.2 0.000198903 0.361133685 ILMN_1784871 FASN NM_004104.4 0.000207395 0.056877471 ILMN_1652716 THEX1 NM_153332.2 0.000207846 0.288484984 ILMN_2050761 EIF4E NM_001968.2 0.00025774 0.385622247 ILMN_1747911 CDC2 NM_001786.2 0.000263771 0.568007158 ILMN_1795340 TMPO NM_001032283.1 0.000264185 0.26979899 ILMN_1780769 TUBB2C NM_006088.5 0.00027646 0.339197541 ILMN_2318430 EIF5 NM_001969.3 0.000283832 0.321862089 ILMN_2051373 NEK2 NM_002497.2 0.000305397 0.631359834 ILMN_1736176 PLK1 NM_005030.3 0.000307467 0.515204021 ILMN_1742238 SET NM_003011.2 0.000327498 0.403070785 ILMN_2159044 PDF NM_022341.1 0.000327825 0.457651002 ILMN_1678669 RRM2 NM_001034.1 0.000346452 0.717686042 ILMN_1721963 MEN1 NM_130801.1 0.000353953 0.671478274 ILMN_1701331 UBE2M NM_003969.3 0.000388326 0.036042361 ILMN_1797693 BRI3BP NM_080626.5 0.000397118 0.239964969 ILMN_2375386 RNPS1 NM_080594.1 0.000413055 0.225058918 ILMN_2390974 DNAJB2 NM_006736.5 0.00042772 11.00703115 ILMN_1727709 GPBAR1 NM_170699.2 0.00045926 1.658706421

Primer number symbol Login ID P value risk ILMN_1792110 C10orf76 NM_024541.2 0.000466902 1.369140571 ILMN_2041327 MRPL37 NM_016491.2 0.000469554 0.053735215 ILMN_1680419 ASB7 NM_024708.2 0.000480501 0.417937044 ILMN_1684873 ARSD NM_001669.2 0.000494954 1.491248628 ILMN_2414399 NME1 NM_000269.2 0.000513696 0.635435354 ILMN_1729368 FZD8 NM_031866.1 0.000538803 1.491192988 ILMN_2354269 FAM164C NM_024643.2 0.000560341 0.580400914 ILMN_2220187 GFPT1 NM_002056.1 0.000563592 0.444796775 ILMN_1693669 WDR79 NM__018081.1 0.000579389 0.432824705 ILMN_2155998 PSMD6 NM_014814.1 0.000581349 0.42069481 ILMN_2116556 LSM5 NM_012322.1 0.00059526 0.244672115 ILMN_1695079 ZNF101 NM_033204.2 0.000608109 0.497757641 ILMN_1661424 THAP6 NM_144721.4 0.000613204 0.505963223 ILMN_1705861 AP1M2 NM_005498.3 0.000620207 0.487707134 ILMN_1788489 HIST1H3F NM_021018.2 0.000652917 0.536377137 ILMN_1740842 SALL2 NM_005407.1 0.000652976 1.634825796 ILMN_1677794 BRCA2 NM_000059.3 0.000653428 0.521624269 ILMN_1712755 LRRC41 NM_006369.4 0.000655598 0.254806566 ILMN_1765532 RDBP NM_002904.5 0.000661777 0.226006062 ILMN_1655734 BXDC5 NM_025065.6 0.000691174 0.321806825 ILMN_1665515 MGC4677 NR_024204.1 0.000696851 0.379310859 ILMN_1652280 FBXO32 NM_058229.2 0.000697182 4.158410279 ILMN_1758067 RGS4 NM_005613.3 0.000731812 1.448721169 ILMN_1677636 COMP NM_000095.2 0.000731916 1.358640294 ILMN_2148796 MND1 NM_032117.2 0.000846675 0.525138393 ILMN_1804090 SLC25A10 NM_012140.3 0.000856602 0.410793488 ILMN__2358914 SLC35C2 NM_015945.10 0.000884827 0.309429971 ILMN_1771385 GBP4 NM_052941.3 0.000892961 0.535961124 ILMN_1780667 WDR51A NM_015426.3 0.000908419 0.509178934 ILMN_1710752 NAPRT1 NM_145201.3 0.000910567 0.122773656 ILMN_1774589 IQCC NM_018134.1 0.000939997 0.39966678 ILMN_1732158 FMO2 NM_001460.2 0.000962173 1.34318011 ILMN_1735453 FAM98A NM_015475.3 0.000981164 0.237917743 ILMN_2265759 SLC2A11 NM_030807.2 0.001008065 1.348676454 ILMN_2190292 UGT8 NM_003360.2 0.001008405 0.335985912 ILMN_2258471 SLC30A5 NM_022902.2 0.001009148 0.353367403 ILMN_1801205 GPNMB NM_001005340.1 0.001012932 1.569348784 ILMN_2224990 HIST1H4J NM_021968.3 0.001040369 0.433804745 ILMN_2396875 IGFBP3 NM_000598.4 0.001056099 1.496136094 ILMN_2079004 MDH2 NM_005918.2 0.001059324 0.098117309 ILMN_2103480 ZNF320 NM_207333.2 0.001120173 0.515683347 ILMN_1756849 HIST1H2AE NM_021052.2 0.001128782 0.55825351 ILMN_1751264 CCDC126 NM_138771.3 0.001131162 0.539458225

Primer number symbol Login ID P value risk ILMN_1670638 PITPNC1 NM_181671.1 0.001156364 1.432837149 ILMN1805404 GRIN1 NM_021569.2 0.001179724 0.646993195 ILMN_1735108 ANKS6 NM_173551.3 0.001190605 0.531136784 ILMN_1710428 CDC2 NM_001786.2 0.00120549 0.30663512 ILMN_1674620 SGCE NM001099400.1 0.001211609 1.454072287 ILMN_1694400 MSR1 NM_138715.2 0.001211833 1.467532171 ILMN_2088847 OTUD5 NM_017602.2 0.00121411 0.378973971 ILMN2160209 TACSTD1 NM_002354.1 0.001214616 0.429796469 ILMN_1803376 AEBP2 NM_153207.3 0.001233925 0.379590397 ILMN_1685431 DZIP1 NM_198968.2 0.001245864 1.355178613 ILMN_2287276 FAM177A1 NM_173607.3 0.001259789 0.515786967 ILMN2241317 FOXK2 NM_004514.3 0.001263375 0.633792641 ILMN_1656192 ZNF704 NM_001033723.1 0.001275697 1.462816957 ILMN_1708105 EZH2 NM_152998.1 0.001284332 0.53646297 ILMN_1778617 TAF9 NM_001015891.1 0.001312314 0.437194069 ILMN_1678423 SPA17 NM_017425.2 0.001322915 0.31024646 ILMN_1735004 C4orf43 NM_018352.2 0.001330807 0.451219666 ILMN_1766264 PI16 NM_153370.2 0.001342722 1.54312461 ILMN_1651429 SELM NM_080430.2 0.001350332 7.120826545 ILMN_1652198 CCM2 NM_001029835.1 0.001360123 1.478634628 ILMN_1651872 UBIAD1 NM_013319.1 0.001363479 0.483856217 ILMN_1747353 KIF27 NM_017576.1 0.001366794 0.305754215 ILMN_1735958 METTL2B NM_018396.2 0.001372067 0.497518932 ILMN_2130441 HLA-H U60319.1 0.001383524 1.309869151 ILMN_2320250 NOL6 NM_022917.4 0.001386545 0.198070845 ILMN_1710170 PPAP2C NM_177526.1 0.001395701 0.643357078 ILMN_1719870 GCUD2 NM_207418.2 0.001456501 0.604998995 ILMN_2181060 CKAP2 NM_001098525.1 0.001459646 0.663378155 ILMN_1669928 ARHGEF16 NM_014448.2 0.001462437 0.435748845 ILMN_2233099 SSRP1 NM_003146.2 0.001462983 0.313100651 ILMN_1788886 TOX NM_014729.2 0.001464337 1.382074069 ILMN_2150894 ALDH1B1 NM_000692.3 0.001471083 0.478898918 ILMN_2148469 RASL11B NM_023940.2 0.001485489 1.301724101 ILMN_1734766 C6orf182 NM_173830.4 0.001504844 0.352005589 ILMN_1811790 FOXS1 NM_004118.3 0.001525515 1.493899582 ILMN_1711543 C14orf169 NM_024644.2 0.001566629 0.378006897 ILMN_1660698 GTPBP8 NM_014170.2 0.001588061 0.40836939 ILMN_1721868 KPNA2 NM_002266.2 0.001617872 0.566719073 ILMN_2344971 FOXM1 NM_202003.1 0.001627322 0.668830609 ILMN_2120340 RUVBL2 NM_006666.1 0.001643506 0.641655709 ILMN1738938 TIMM8B NM_012459.1 0.001649996 0.507398524 ILMN_1718387 LOR NM_000427.2 0.001653634 1.304994604 ILMN_2388517 MTERFD3 NM_001033050.1 0.001661168 0.298295273

Primer number symbol Login ID P value risk ILMN_1795063 ZADH2 NM_175907.3 0.001673092 1.385143264 ILMN_1695579 CIT NM_007174.1 0.001678569 0.631478718 ILMN_1767448 LHFP NM_005780.2 0.001686324 1.649522164 ILMN_1712803 CCNB1 NM_031966.2 0.001729899 0.698597294 ILMN_1715583 BOP1 NM_015201.3 0.001745263 0.427680383 ILMN_1685343 NUPL1 NM_001008565.1 0.001752276 0.531577742 ILMN_1790781 DHRS13 NM_144683.3 0.00180559 0.373621553 ILMN_2152387 DOCK7 NM_033407.2 0.001860415 0.372834463 ILMN_2223836 CHORDC1 NM_012124.1 0.001888776 0.314995736 ILMN_1775925 HIST1H2BI NM_003525.2 0.00191329 0.215094156 ILMN_2310909 ATP2A3 NM_174955.1 0.001920605 0.640714872 ILMN_1799999 LRRCC1 NM_033402.3 0.001933642 0.310223122 ILMN_1765258 HLA-E NM_005516.4 0.001962169 0.41448045 ILMN_1693604 GRM2 NM_000839.2 0.001964674 0.605639742 ILMN_2396020 DUSP6 NM_001946.2 0.001989538 0.440365228 ILMN_2215370 WWP1 NM_007013.3 0.002004931 0.410021251 ILMN_1732127 RBKS NM_022128.1 0.002028283 0.519499482 ILMN_1709451 TFPT NM_013342.2 0.00203197 2.444929835 ILMN_1753467 SAMD4B NM_018028.2 0.002084896 1.944528201 ILMN_1797893 PFAAP5 NM_014887.1 0.002089509 1.780690735 ILMN_2413898 MCM10 NM_018518.3 0.002098445 0.746035804 ILMN_1689086 CTSC NM_001814.2 0.002101771 0.357197864 ILMN_2363668 YIF1B NM_001039673.1 0.002144846 0.395212683 ILMN_1737195 CENPK NM_022145.3 0.002181123 0.652279 ILMN_1749789 HIST1H1D NM_005320.2 0.002196903 0.45086036 ILMN_1760153 GATA5 NM_080473.3 0.00224015 1.333301072 ILMN_2369018 EVI2A NM_014210.2 0.002248908 1.53070581 ILMN_2294653 PDE5A NM_033437.2 0.00226466 1.325372841 ILMN_1742307 MEST NM_177524.1 0.002265982 0.384915559 ILMN_2207865 HIST1H3I NM_003533.2 0.002313157 0.390719052 ILMN_2061043 CD48 NM_001778.2 0.002343934 1.4403949 ILMN_2307656 AGTRAP NM_001040196.1 0.002353483 0.476416601 ILMN_1737709 RPL10L NM_080746.2 0.002355167 0.592421921 ILMN_1777156 GTPBP3 NM_032620.1 0.002378577 0.458656308 ILMN_2058141 HMGN2 NM_005517.3 0.00240793 0.396669549 ILMN1700413 MAFF NM_152878.1 0.002420337 0.473434773 ILMN_1688755 AAK1 NM_014911.2 0.0024249 1.443142271 ILMN_1656415 CDKN2C NM_078626.2 0.002445334 1.363082198 ILMN_1773080 OAZ1 NM_004152.2 0.002463203 0.006652434 ILMN_1655052 TRNT1 NM_016000.2 0.002469548 0.453240253 ILMN_1763491 CKMT1B NM_020990.3 0.002479888 0.696617712 ILMN_2349459 BIRC5 NM_001012271.1 0.002493781 0.632059133 ILMN_1693597 ZNF287 NM_020653.1 0.002546359 1.321302822

Primer number symbol Login ID P value risk ILMN_1791149 ARL6IP4 NM_001002252.1 0.002559255 0.445204912 ILMN_2191634 RPL37 NM_000997.3 0.002566623 0.439638182 ILMN_1692511 TMEM106C NM_024056.2 0.002584001 0.410086273 ILMN_2336335 40245 NM_006231.2 0.002594922 1.492701021 ILMN_1670903 NAT2 NM_000015.2 0.002621283 0.394117706 ILMN2350183 ST5 NM_213618.1 0.00264123 1.71088169 ILMN_1806473 BEX5 NM_001012978.2 0.00264497 1.405290668 ILMN1745108 ADAD2 NM_139174.2 0.002650355 0.564020967 ILMN_2208455 DDHD1 NM_030637.1 0.002666187 0.62040113 ILMN2289381 DKK3 NM_015881.5 0.002677686 1.273699695 ILMN_2093500 ZBED5 NM_021211.2 0.002686929 1.593462221 ILMN_1676215 DLG2 NM_001364.2 0.002687616 1.393751447 ILMN_1746435 HIST1H1E NM_005321.2 0.002709568 0.426535974 ILMN_1681757 FAM80B NM_020734.1 0.002716862 1.385585016 ILMN_1814282 ISG20L1 NM_022767.2 0.002720111 0.401067743 ILMN_1695107 IL20RA NM_014432.2 0.002727132 0.391212161 ILMN_1704261 RANGRF NM_016492.3 0.002732176 0.561395184 ILMN_1742544 MEF2C NM_002397.2 0.002763958 1.634742949 ILMN_1800420 RNF214 NM_207343.2 0.002818409 0.441702219 ILMN_2115696 USP42 NM_032172.2 0.002826683 1.254016918 ILMN_2369104 TRAPPC6B NM_177452.3 0.002836961 0.537255223 ILMN_1811426 TMTC1 NM_175861.2 0.002846937 1.449228612 ILMN_1679641 FAM120B NM_032448.1 0.00286719 0.700685077 ILMN_2191436 POLA1 NM_016937.3 0.002880446 0.291792953 ILMN_1708160 KPNA2 NM_002266.2 0.002920159 0.671481421 ILMN_1752249 FAM38A NM_014745.1 0.002968991 5.849847658 ILMN_2414027 CKLF NM_001040138.1 0.003007182 0.536219708 ILMN_1748147 MTO1 NM_133645.1 0.003021874 0.435348875 ILMN_1688231 TREM1 NM_018643.2 0.003038524 0.404373704 ILMN_2071826 RNF152 NM_173557.1 0.003053367 1.396524052 ILMN_1720542 POLR2I NM_006233.4 0.00316402 0.710946957 ILMN_1718334 ITPA NM_033453.2 0.003165015 0.441320307 ILMN_1731374 CPE NM_001873.1 0.003168171 1.387565375 ILMN_2099045 KIAA1524 NM_020890.1 0.003184317 0.397921546 ILMN_1774350 MYOZ3 NM_133371.2 0.003202754 1.362439801 ILMN_2395926 MANBAL NM_022077.3 0.003220826 0.446106468 ILMN_1814002 TEAD3 NM_003214.3 0.00322364 0.508063459 ILMN_2351916 EX01 NM_006027.3 0.00323852 0.621409496 ILMN_1785005 NCF4 NM_013416.2 0.003244984 1.852600594 ILMN_2082810 BRD7 NM_013263.2 0.003247679 0.445190505 ILMN_1702858 ADHFE1 NM_144650.2 0.003259448 1.522440502 ILMN_1815385 SMAD9 NM_005905.3 0.003266862 1.486381567 ILMN_1699665 CLIC6 NM_053277.1 0.003268017 1.394934339

Primer number symbol Login ID P value risk ILMN_1672660 MBP NM_001025100.1 0.003308632 1.307617077 ILMN_1710495 PAPLN NM_173462.3 0.003317357 1.555971585 ILMN_1788955 PDLIM1 NM_020992.2 0.003335768 0.36434918 ILMN_1750130 GSPT1 NM_002094.2 0.003432206 0.561549921 ILMN_1715175 MET NM_000245.2 0.00344409 0.408579482 ILMN_1688041 TMEM53 NM_024587.2 0.003458129 0.460287691 ILMN_1693333 TMEM19 NM_018279.3 0.003507351 0.36821929 ILMN_1688848 TMEM44 NM_138399.3 0.003510688 0.444391423 ILMN_2379527 ELM01 NM_014800.9 0.003566733 1.44357207 ILMN_1729713 RAB23 NM_183227.1 0.003569889 1.290055449 ILMN_1717262 PROCR NM_006404.3 0.003588508 0.481142471 ILMN_1722829 HLF NM_002126.4 0.003592205 1.387440105 ILMN_1653165 AAMP NM_001087.3 0.003625108 0.51315174 ILMN_1652826 LRRC17 NM_005824.1 0.0036285 1.338749637 ILMN_1653200 SLC22A17 NM_020372.2 0.003656112 1.452161875 ILMN_2076250 GPBP1L1 NM_021639.3 0.003660253 0.525101894 ILMN_1731610 ABLIM1 NM_006720.3 0.00367075 1.420084637 ILMN_1653001 CABLES1 NM_138375.1 0.003735355 0.421734171 ILMN_1670609 ATOX1 NM_004045.3 0.003736891 4.196917472 ILMN_1714197 ACSS2 NM_139274.1 0.003759326 0.144997669 ILMN_2367070 ACOT9 NM_001033583.2 0.003773424 0.439029956 ILMN_1757406 HIST1H1C NM_005319.3 0.00377794 0.545763401 ILMN_1815010 RNF141 NM_016422.3 0.003799608 0.499957309 ILMN_1687589 CPT1A NM_001876.2 0.003816375 0.556334372 ILMN_1702265 HDHD2 NM_032124.4 0.003820466 0.255780221 ILMN_1776577 DSCC1 NM_024094.2 0.003827106 0.445375329 ILMN_1680692 NUCKS1 NM_022731.2 0.003830411 0.352050942 ILMN_2301083 UBE2C NM_181803.1 0.00384705 0.746466822 ILMN_1660636 WWOX NM_130844.1 0.003850247 0.389587873 ILMN_2252408 CNPY4 NM_152755.1 0.003872081 1.926411669 ILMN_2122374 FAM49B NM_016623.3 0.003874185 0.475549021 ILMN_1679809 GSTP1 NM_000852.2 0.00390375 12.13414752 ILMN_1739645 ANLN NM_018685.2 0.003925781 0.565399848 ILMN_1804419 LRMP NM_006152.2 0.003992031 1.293737566 ILMN_2330410 EIF3C NM_003752.3 0.004011169 0.321444331 ILMN_1696380 GHRL NM_016362.2 0.004018356 1.26994829 ILMN_1787280 C1orf135 NM_024037.1 0.004058179 0.650335614 ILMN_1736178 AEBP1 NM_001129.3 0.004064418 2.22040315 ILMN_1801939 CCNB2 NM_004701.2 0.004065318 0.738524159 ILMN_1682375 ATPBD3 NM_145232.2 0.004071346 0.47285411 ILMN_1657701 TMEM137 XR_017971.1 0.004078655 0.178899275 ILMN_1662419 COX7A1 NM_001864.2 0.004102612 1.701574095 ILMN_1708041 PLEKHF1 NM_024310.4 0.004131276 1.32143517

Primer number symbol Login ID P value risk ILMN_1667641 ACACA NM_198834.1 0.004154722 1.323720243 ILMN_1682675 TWF1 NM_002822.3 0.004217274 0.474590351 ILMN_2123402 TMEM4 NM_014255.4 0.00423593 0.426349629 ILMN_1720484 CRTAP NM_006371.3 0.004250147 1.390315752 ILMN_1756982 CLIC1 NM_001288.4 0.004251433 0.262265066 ILMN_2315964 PSRC1 NM_001032290.1 0.004265342 0.446786367 ILMN_1738704 TRIM26 NM_003449.3 0.004269347 0.598757026 ILMN_1713178 FAM116A XM_001132771.1 0.004275545 0.563919761 ILMN_1814856 C9orf7 NM_017586.1 0.004277418 0.424020863 ILMN_1755504 CALCOCO2 NM_005831.3 0.004300376 1.498460692 ILMN_1764694 ZFP14 NM_020917.1 0.004300529 1.492032427 ILMN_1718265 ATG5 NM_004849.2 0.004309361 0.314797866 ILMN_1764850 HPCAL1 NM_134421.1 0.004323109 0.564769594 ILMN_1677652 PREX2 NM_024870.2 0.004337807 1.291465793 ILMN_2362293 FBXO38 NM_205836.1 0.004344345 0.30957276 ILMN_2184231 CHRDL1 NM_145234.2 0.004355928 1.405989697 ILMN_1674337 FKBP2 NM_057092.1 0.004365515 0.552251971 ILMN_1673380 GNG12 NM_018841.4 0.004429613 2.220091452 ILMN_1730347 CCDC115 NM_032357.2 0.004442253 1.299753115 ILMN_1752589 TMEM183A NM_138391.4 0.004454617 0.582347893 ILMN_1692790 ITGB3BP NM_014288.3 0.004456606 0.283335019 ILMN_1680626 PDIA6 NM_005742.2 0.004486813 0.438758243 ILMN_1789040 SLITRK5 NM__015567.1 0.004487421 1.58800473 ILMN_2221046 GM2A NM_000405.3 0.004499557 1.343681008 ILMN_2392818 RTKN NM_D33046.2 0.004504209 0.380099265 ILMN_1691559 ELF2 NM_006874.2 0.004539554 1.445911237 ILMN_2120965 NPAT NM_002519.1 0.004608556 1.53213071 ILMN_1761772 NUP155 NM_153485.1 0.004609259 0.787766929 ILMN_1768969 LBR NM_194442.1 0.004612078 0.381969617 ILMN_1669931 TM9SF3 NM_020123.2 0.004639769 0.455109316 ILMN_1731194 STRAP NM_007178.3 0.004663374 0.486661013 ILMN_1665717 EIF2S3 NM_001415.3 0.00466426 0.607788466 ILMN_2076567 UBE2V2 NM_003350.2 0.004673328 0.376889407 ILMN_1815570 HOXA6 NM_024014.2 0.004677877 1.315082473 ILMN_1704943 ATPBD1C NM_016301.2 0.004692424 0.45831831 ILMN_1681304 PAN3 NM_175854.5 0.004694593 0.336691298 ILMN_1754842 DLGAP4 NM_014902.3 0.004695897 1.500771594 ILMN_2397347 SEMG1 NM_198139.1 0.004704468 0.504369353 ILMN_1766983 FBXW11 NM_033644.2 0.004733209 1.373759174 ILMN_1715607 CHMP4A NM_014169.2 0.004813934 0.599264497 ILMN_1657148 C19orf23 NM_152480.1 0.004839756 0.777157317 ILMN_1749213 SDF2L1 NM_022044.2 0.004874196 0.117041269 ILMN_1664761 TMEM138 NM_016464.3 0.004884175 1.768386882

Primer number symbol Login ID P value risk ILMN_1782403 PRR11 NM_018304.2 0.004895172 0.692001418 ILMN_1749583 KIAA1285 NM_015694.2 0.004895535 0.485704875 ILMN_2294274 S100PBP NM_022753.2 0.004902704 0.348252651 ILMN_2089977 FKBP9L NM_182827.1 0.004925827 1.35715531 ILMN_1708143 FAM127A NM_001078171.1 0.004940424 1.536857038 ILMN_1687947 HIST1H2BE NM_003523.2 0.004945146 0.60890965 ILMN_1790741 RNF126 NM_194460.1 0.004999963 0.403652191 ILMN_2084391 RAD18 NM_020165.2 0.005021514 0.478869988 ILMN_1700975 ENSA NM_207168.1 0.005023952 0.597039464 ILMN_1758529 P2RX1 NM_002558.2 0.005040437 1.331690952 ILMN_1653824 LAMC2 NM_018891.1 0.005053334 0.510397105 ILMN_1673673 PBK NM_018492.2 0.005072448 0.502353808 ILMN_2188451 HIST1H2AH NM_080596.1 0.005091975 0.56869136 ILMN_1729430 FBXO18 NM_032807.3 0.005096865 0.409638443 ILMN_2145670 TNC NM_002160.2 0.005099349 0.44945639 ILMN_1799113 CCDC41 NM_016122.2 0.005124666 0.331541717 ILMN_1694177 PCNA NM_182649.1 0.005131856 0.431824254 ILMN_2365176 ALDH8A1 NM_022568.2 0.0051498 0.578765996 ILMN_1703791 ANXA7 NM_004034.1 0.005194134 0.520011299 ILMN_1653432 HNRPDL NR_003249.1 0.00519902 1.962006317 ILMN_1711470 UBE2T NM_014176.2 0.005212905 0.70825341 ILMN_1672876 MFI2 NM_05929.4 0.005216636 0.688790743 ILMN_1803956 BOC NM_033254.2 0.005222405 1.56055792 ILMN_1793959 ADPGK NM_031284.3 0.005222668 0.622209617 ILMN_2141118 C15orf59 NM_001039614.1 0.005232281 1.288389631 ILMN_1740265 ACOT7 NM_181864.2 0.005276176 0.42513968 ILMN_1705515 UPF3A NM_080687.1 0.005335596 0.555924796 ILMN_1747870 CD3EAP NM_012099.1 0.005335819 0.431594263 ILMN_1662935 C1QTNF7 NM_031911.3 0.005340141 1.400132792 ILMN_2408796 C190rf28 NM_174983.3 0.005377619 0.535188081 ILMN_1808748 CLCN6 NM_001286.2 0.005389385 0.448798804 ILMN_2347999 IFNAR2 NM_207585.1 0.00540236 0.394741019 ILMN_1759184 C190rf48 NM_199250.1 0.005420614 0.262738545 ILMN_2402392 COL8A1 NM_001850.3 0.005441026 1.506073801 ILMN_1670542 AK2 NM_001625.2 0.005456193 0.4642596 ILMN_1815306 AP2A1 NM_014203.2 0.005479068 0.45782699 ILMN_1665982 AKTIP NM_022476.2 0.005552907 1.426606686 ILMN_1754476 TRIM15 NM_033229.2 0.005599962 0.672365812 ILMN_1715789 DOCK1 NM_001380.3 0.005637866 1.290809728 ILMN_2140207 ATPBD4 NM_080650.2 0.005687417 0.465173775 ILMN_1707257 HIST1H3J NM_003535.2 0.00569548 0.411868035 ILMN_2330341 TCEAL4 NM_024863.4 0.005697718 1.905792982 ILMN_2371964 MRPS12 NM_021107.1 0.005734333 0.338999466

Primer number symbol Login ID P value risk ILMN_1793888 SERPIN B5 NM_002639.3 0.005756119 0.759749969 ILMN_1715616 PPIL5 NM_203467.1 0.005771905 0.454389155 ILMN_1702526 C17orf48 NM_020233.4 0.005799246 1.355222195 ILMN_1739076 HIST1H2BO NM_003527.4 0.005801406 0.73401284 ILMN_2075714 ZNF284 NM_001037813.2 0.005824027 1.411449642 ILMN_1814151 AGR2 NM_006408.2 0.005831268 0.504940913 ILMN_1738684 NRXN2 NM_138734.1 0.005836232 1.418214009 ILMN_2065022 KIAA0672 NM_014859.4 0.00584505 1.279720801 ILMN_2402168 EXOSC10 NM_001001998.1 0.005845299 0.518518591 ILMN_1803570 BRI3BP NM_080626.5 0.00584885 0.795690999 ILMN_2103362 ARHGAP27 NM_199282.1 0.005891206 0.487395295 ILMN_1731048 TLR1 NM_003263.3 0.005909894 0.354107909 ILMN_1813295 LMO3 NM_018640.3 0.00592348 1.417834019 ILMN_1676058 MAGOHB NM_018048.3 0.005932808 0.466547235 ILMN_2255133 BCL11A BCL11A 0.005943127 0.464319877 ILMN_2311537 HMGA1 NM_145902.1 0.005946164 0.801243797 ILMN_1718853 UQCRC2 NM_003366.2 0.005980146 0.513524684 ILMN_1776845 HIST1H3A NM_003529.2 0.005985544 0.710949575 ILMN_1672122 PH-4 NM_177938.2 0.005995002 0.5034317 ILMN_1651229 IP013 NM_014652.2 0.006001369 2.321489493 ILMN_2217661 SREBF2 NM_004599.2 0.006020596 0.42538648 ILMN_2115340 HIST2H4A NM_003548.2 0.006051534 0.723953441 ILMN_1662140 SGPP2 NM_152386.2 0.006053142 0.735761893 ILMN_2362368 U2AF1 NM_001025203.1 0.006055125 0.543969023 ILMN_1710070 PCSK6 NM_138320.1 0.00611501 0.72158868 ILMN_2358783 ASB3 NM_016115.3 0.006128461 0.496862366 ILMN_2407464 FASTK NM_006712.3 0.006201308 0.537689921 ILMN_2382990 HK1 NM_033498.1 0.006216229 2.353179628 ILMN_2143685 CLDN7 NM_001307.4 0.00624499 0.7855261 ILMN_1726108 LASS2 NM_181746.2 0.006250473 0.422395144 ILMN_1734867 NR2C1 NM_003297.1 0.006253786 0.575169323 ILMN_1788180 RAB13 NM_002870.2 0.006265118 0.448550857 ILMN_1720595 MDGA1 NM_153487.3 0.006287087 1.594869983 ILMN_2162358 ZNF597 NM_152457.1 0.006307603 1.269525875 ILMN_1717393 PTCHD1 NM_173495.2 0.006308428 1.365299066 ILMN_1688033 HPS5 NM_181507.1 0.006328446 1.350346589 ILMN_1664815 ELK4 NM_001973.2 0.006330513 0.648779128 ILMN_2388070 TMEM44 NM_138399.3 0.006349048 0.462595584 ILMN_1666096 ACSL3 NM_004457.3 0.006367395 1.470484977 ILMN_1717757 CALML4 NM_001031733.2 0.006395212 0.732505749 ILMN_2116827 RGPD1 NM_001024457.1 0.006400663 1.273196499 ILMN_1684647 ILKAP NM_030768.2 0.006420595 0.520137067 ILMN_1795507 ABCA6 NM_080284.2 0.00645301 1.266389585

Primer number symbol Login ID P value risk ILMN_1726030 GPX7 NM_015696.3 0.006505417 1.319011509 ILMN_2336595 ACSS2 NM_018677.2 0.006519276 0.377268086 ILMN_1653251 HIST1H1B NM_005322.2 0.006533608 0.61504306 ILMN_2250923 FOXP1 NM_032682.4 0.006544389 0.435456884 ILMN_1694759 C19orf42 NM_024104.3 0.006574444 0.480847866 ILMN_2230025 PDLIM3 NM_014476.1 0.006617224 1.749382931 ILMN_1812970 RWDD1 NM_016104.2 0.006651454 1.274265391 ILMN_1733559 LOC100008589 NR_003287.1 0.006655633 0.05494194 ILMN_2214278 ANKRD32 NM_032290.2 0.006672263 0.407002498 ILMN_2364928 APBA2BP NM_031231.3 0.006674687 0.476662735 ILMN_2368721 CENPM NM_024053.3 0.006681262 0.775688614 ILMN_2042651 EVI2B NM_006495.3 0.006709516 0.398923979 ILMN_1757536 USP40 NM_018218.2 0.006719019 0.443606755 ILMN_1743579 WDR4 NM_033661.3 0.006719975 0.586652104 ILMN_1794017 SERTAD1 NM_013376.3 0.006766851 1.477182017 ILMN_2192683 DHX37 NM_032656.2 0.006803347 0.144674391 ILMN_2148452 BCAS2 NM_005872.2 0.00688614 0.394539719 ILMN_1805778 RBM12B NM_203390.2 0.006906573 1.534168009 ILMN_1658821 SAMD1 NM_138352.1 0.006914193 0.710097017 ILMN_2072357 IRF6 NM_006147.2 0.006953237 0.502557868 ILMN_1740508 KCNMA1 NM_001014797.1 0.006971733 1.4569512 ILMN_2401779 FAM102A NM_001035254.1 0.007011925 0.251786364 ILMN_2330570 LEPR NM_002303.3 0.007074066 1.44870954 ILMN_1675106 YIPF2 NM_024029.3 0.007074451 0.454951256 ILMN_1784367 HSPD1 NM_002156.4 0.007116294 0.729230188 ILMN_1798254 ACTR10 NM_018477.2 0.00713915 0.456078802 ILMN_2061950 RABGAP1 NM_012197.2 0.007143416 1.909688817 ILMN_1657836 PLEKHG2 NM_022835.1 0.007225484 1.409139505 ILMN_2073307 IL10 NM_000572.2 0.007235338 0.581576619 ILMN_1669023 FHL5 NM_020482.3 0.007238126 1.314020057 ILMN_2413251 EWSR1 NM_005243.2 0.007243401 0.497456343 ILMN_1692779 PRPF39 NM_017922.2 0.007254878 0.410654488 ILMN_1803338 CCDC80 NM_199511.1 0.007368666 1.892126506 ILMN_2316918 PANK1 NM_148978.1 0.007397188 0.417679556 ILMN_1781400 SLC7A2 NM_001008539.2 0.007413353 1.515745654 ILMN_1799289 MRPL55 NM_181454.1 0.007420628 0.461570361 ILMN_2410924 PLOD2 NM_000935.2 0.007437692 1.876219594 ILMN_1684931 GPR119 NM_178471.1 0.007438219 0.463748761 ILMN_2138589 MERTK NM_006343.2 0.007438466 1.473780814 ILMN_1671557 PHLDA2 NM_003311.3 0.00745336 0.620891122 ILMN_1809101 STEAP2 NM_152999.3 0.007473192 1.405229708 ILMN_2381037 LIMS1 NM_004987.3 0.007479728 0.506576202 ILMN_1723522 APOLD1 NM_030817.1 0.007500182 1.273846061

Primer number symbol Login ID P value risk ILMN_2292178 CLEC12A NM_201623.2 O.007513142 1.293618589 ILMN_2294684 CEP170 NM_014812.2 O.007567842 0.492080122 ILMN_2331163 CUL4A NM_003589.2 O.007580851 O.677020715 ILMN_2209163 CHD6 NM_032221.3 O.007605645 O.532521441 ILMN_1777263 MEOX2 NM_005924.4 O.007648662 1.344671457 ILMN_1688666 HIST1H2BH NM_003524.2 0.007688938 O.679188681 ILMN_2064926 ITFG1 NM_030790.3 O.007689192 O.430133813 ILMN_1812795 RUNX1T1 NM_175636.1 O.007693535 1.234743018 ILMN_1783908 B3GNT9 NM_033309.2 0.007740725 0.548726198 ILMN_1689438 BTRC NM_033637.2 O.007752648 1.397535008 ILMN_1687652 TGFB3 NM_003239.1 O.007770746 1.352457691 ILMN_1695025 CD2 NM_OO1767.3 O.007797265 1.313736386 ILMN_1749846 OMD NM_005014.1 O.007801275 1.462435251 ILMN_1807945 ANP32A NM_006305.2 0.0078657 0.54232047 ILMN_166491O RPSA NM_001012321.1 O.007938497 O.537155609 ILMN_2407605 GIYD2 NM_024044.2 O.007944606 O.531217635 ILMN_2152095 RNASEN NM_013235.3 O.007987447 O.458218869 ILMN_1678464 DCLRE1C NM_001033858.1 O.007991392 O.619351961 ILMN_1730529 CAB39L NM_001079670.1 O.008004106 O.465761281 ILMN_1809285 DCP1A NM_018403.4 O.008063544 O.499569903 ILMN_1699440 ZBTB47 NM_145166.2 O.008069873 1.300721326 ILMN_1774083 TRIAP1 NM_016399.2 O.008084101 O.452093496 ILMN_1796523 FNIP1 NM_133372.2 0.008101041 1.251935595 ILMN_1742379 IFT122 NM_052989.1 0.008107069 O.585299967 ILMN_1798581 MCM8 NM_032485.4 O.008124758 O.39895875 ILMN_2045994 SEPW1 NM_003009.2 O.008180544 2.170610772 ILMN_2111237 MN1 NM_002430.2 O.008181064 1.480543494 ILMN_1727558 MRPL27 NM_148571.1 O.008216167 O.49290801 ILMN_1713613 PIAS2 NM_173206.2 O.008218756 O.533318798 ILMN_2207720 ITM2B NM_021999.3 O.008268371 0.409229979 ILMN_1778059 CASP4 NM_033306.2 0.008299631 O.352535107 ILMN_2151281 GABARAPL1 NM_031412.2 0.00830592 1.40344079 ILMN_2227368 SELT NM_016275.3 O.008323563 0.629151938 ILMN_1755222 C9orf82 NM_024828.2 0.008376736 0.537845874 ILMN_1670272 LRP10 NM_014045.3 O.008387951 0.488461901 ILMN_1750044 ZNHIT3 NM_004773.2 O.008414785 0.596203697 ILMN_1801899 PLEC1 NM_201380.2 O.008415501 O.788422978 ILMN_1667707 SPCS3 NM_021928.1 O.008416213 0.333306413 ILMN_1784459 MMP3 NM_002422.3 O.008424899 0.743845984 ILMN_1715613 TAOK2 NM_004783.2 O.008428815 O.577192898 ILMN_1783170 ING3 NM_198267.1 O.008464107 O.45594537 ILMN_2343332 TAF9 NM_001015891.1 O.008467944 O.502769469 ILMN_1746426 TOMM70A NM_014820.3 O.008511501 0.451400009

Primer number symbol Login ID P value risk ILMN_1708164 EIF3A NM_003750.2 0.0085147 0.273046283 ILMN_2319919 MAGEA2 NM_175743.1 0.008522299 0.564708047 ILMN_1788166 TTK NM_003318.3 0.008532079 0.6260955 ILMN_1694731 CLCN7 NM_001287.3 0.008552366 0.047428986 ILMN_2189037 WDR52 NM_018338.2 0.008579091 0.442426039 ILMN_1654411 CCL18 NM_002988.2 0.008582303 0.534771494 ILMN_1666372 ATP5H NM_006356.2 0.00858678 0.339181294 ILMN_1663220 MRPL22 NM_014180.2 0.008657699 0.682213605 ILMN_1736689 PC NM_001040716.1 0.008686562 0.491364567 ILMN_1702609 B3GNT5 NM_032047.4 0.008699695 0.35924426 ILMN_2126399 psiTPTE22 EF535614.1 0.008701089 0.508410979 ILMN_1714438 MUTYH NM_001048172.1 0.008708257 1.452044111 ILMN_1697117 TBP NM_003194.3 0.008717257 1.375754246 ILMN_2160476 CCL22 NM_002990.3 0.00873155 0.626816124 ILMN_2405156 PPAP2C NM_177543.1 0.008744942 0.486658323 ILMN_2357976 BAT1 NM_004640.5 0.008748889 0.39869983 ILMN_1686804 CCRK NM_012119.3 0.008781738 1.379683611 ILMN_1735908 UTP15 NM_032175.2 0.008795849 0.455716604 ILMN_1739496 PRRX1 NM_006902.3 0.008810631 1.352203092 ILMN_2215631 OTUD6B NM_016023.2 0.008824788 0.318170009 ILMN_1676449 SLIT2 NM_004787.1 0.008825626 1.443309256 ILMN_1717982 BZW1 NM_014670.2 0.008843056 0.496720206 ILMN_2089902 NUS1 NM_138459.3 0.008924061 0.426437437 ILMN2172269 TMEM183B NM_001079809.1 0.008947562 0.531376285 ILMN_1750144 C3orf19 NM_016474.4 0.008998487 1.224881041 ILMN_1686043 FAM164C NM_024643.2 0.009007654 0.524954198 ILMN_2342793 FBXW8 NM_153348.2 0.009010131 0.526747165 ILMN_1684554 COL16A1 NM_001856.3 0.009207494 2.157593197 ILMN_2198878 INPP4B NM_003866.1 0.009224731 1.355483357 ILMN_2183610 SERAC1 NM_032861.2 0.009237213 0.652929929 ILMN_1735827 NISCH NM_007184.3 0.009267976 3.244972339 ILMN_2408815 NAP1L1 NM_139207.1 0.009273677 0.466078202 ILMN_1736154 ProSAPiP1 NM_014731.2 0.009297408 1.371836507 ILMN_1660043 UBXN11 NM_145345.2 0.009301234 0.414923686 ILMN_1723087 MDK NM_002391.3 0.009357376 0.477925138 ILMN_2405190 VAPA NM_003574.5 0.009380725 0.642869229 ILMN_2365549 BRPF1 NM_004634.2 0.009418423 0.030900528 ILMN_1688637 TMEM198 NM_001005209.1 0.009454295 0.53441861 ILMN_2381753 G3BP2 NM_012297.3 0.009454643 0.528560031 ILMN_2119945 NDUFB3 NM_002491.1 0.009512793 0.605900377 ILMN_2395913 ARHGAP11A NM_199357.1 0.009513189 0.52537859 ILMN_1799105 COL17A1 NM_000494.3 0.009521418 0.569237452 ILMN_1786707 C19orf63 NM_175063.4 0.009549933 0.705859833

Primer number symbol Login ID P value risk ILMN_1695962 SLC12A9 NM_020246.2 0.00955532 0.632986434 ILMN_2213247 SPCS2 NM_014752.1 0.009568632 0.468103464 ILMN1722016 LY6G5C NM_001002849.1 0.009572932 1.337326232 ILMN_1809141 ING4 NM_198287.1 0.009584963 0.47920737 ILMN_1682332 GYPC NM_016815.2 0.009633083 1.887056171 ILMN_1761363 VAMP4 NM_003762.3 0.009646552 0.476950334 ILMN_1785179 UBE2G2 NM_003343.4 0.009701867 0.617105146 ILMN_1763000 ADAP2 NM_018404.2 0.009748952 1.681799633 ILMN_1716224 STARD4 NM_139164.1 0.009783456 0.489406267 ILMN_1739587 UTY NM_007125.3 0.009816309 0.439298478 ILMN_1680643 KIAA1333 NM_017769.2 0.009820691 0.460192694 ILMN_2407879 SORBS2 NM_003603.4 0.009831185 1.548113873 ILMN_2275533 DIAPH3 NM_030932.3 0.009833778 0.686553443 ILMN_1773645 GMPPB NM_021971.1 0.009837559 0.446130182 ILMN_1764091 R3HDM2 NM_014925.2 0.009843159 0.685043284 ILMN_2400500 LASS2 NM_181746.2 0.009847868 0.442253821 ILMN_2334350 BTBD3 NM_014962.2 0.00985416 0.459870001 ILMN_1660602 C1orf43 NM_015449.2 0.009874084 0.617987793 ILMN_2170353 PTPLB NM_198402.2 0.009918567 0.400585431 ILMN_1794492 HOXC6 NM_153693.3 0.00992158 0.721974031 ILMN_2364357 RPS6KB2 NM_003952.2 0.009937711 0.244179599 ILMN_2147503 ALG13 NM_018466.3 0.009946793 0.352771071 ILMN_2177460 AQR NM_014691.2 0.009953961 0.404663806 ILMN_1668525 NR3C1 NM_001018076.1 0.009992789 0.617215381 ILMN_2258543 PRDM2 NM_012231.3 0.009994008 1.364891701

Example 6: Identification of reference genes for self-normalization

One of the ways to reduce the number of prognostic genes found in the clinical domain is to self-normalize in each case, since it is not possible to normalize against the entire group of patients at once. Currently, real-time QRT-PCR (quantitative real-time reverse transcription polymerase chain reaction) is widely used to measure gene expression levels, but when using QTR-PCR, all genes in humans cannot be measured for quantile normalization, And there is the problem that the real-time QRT-PCR signal generated is significantly lower when using old paraffin blocks instead of using new samples.

Therefore, in order to reliably use the identified gastric cancer prognostic genes in the clinical field, the present inventors endeavored to identify reference genes for self-normalization of the measured gene expression levels. Therefore, by analyzing the gene expression level data measured with WG-DASL in Example 3, 50 reference genes that did not have prognostic features and showed minimal changes in each of the different cases were identified, and the results are shown in Table 4. result. The combination of one or more genes among the 50 reference genes listed in Table 4 can be used to normalize the expression levels of gastric cancer prognostic genes.

Table 4

Probe No. Login ID symbol ILMN_2403446 NM_007011.5 ABHD2 ILMN_1708502 NM_014423.3 AFF4 ILMN_1780806 NM_025190.3 ANKRD36B ILMN_2415467 NM_080550.2 AP1GBP1 ILMN_1722066 NM_018120.3 ARMC1 ILMN_2352934 NM_004318.2 ASPH ILMN_1808163 NM_022338.2 C11orf24 ILMN_1693431 NM_153218.1 C13orf31 ILMN_1733288 NM_016546.1 C1RL ILMN_2202940 NM_020244.2 CHPT1 ILMN_2188533 NT_007592.15 CICK0721Q.1 ILMN_1795754 NM_001289.4 CLIC2 ILMN_2373779 NM_198189.2 COPS8 ILMN_1796180 NM_021117.2 CRY2 ILMN_1651499 NM_020462.1 ERGIC1 ILMN_1751425 NM_024896.2 ERMP1 ILMN_1712095 NM_005938.2 FOXO4 ILMN_1747305 NM_175571.2 GIMAP8 ILMN_1737308 NM_002064.1 GLRX ILMN_2168215 NM_016153.1 HSFX1 ILMN_1789018 NM_012218.2 ILF3 ILMN_1809141 NM_016162.2 ING4 ILMN_1807767 NM_014615.1 KIAA0182 ILMN_1776963 NM_006816.1 LMAN2 ILMN_1743583 NM_130473.1 MADD

Probe No. Login ID symbol ILMN_1774844 NM_032960.2 MAPKAPK2 ILMN_1761858 NM_033290.2 MID1 ILMN_1670801 NM_000254.1 MTR ILMN_1780937 NM_025128.3 MUS81 ILMN_1784113 NM_020378.2 NAT14 ILMN_2147133 NM_173638.2 NBPF15 ILMN_2361185 NM_024878.1 PHF20L1 ILMN_1704529 NM_021130.3 PPIA ILMN_2357577 NM_006251.5 PRKAA1 ILMN_2353202 NM_152882.2 PTK7 ILMN_1813753 NM_002825.5 PTN ILMN_1677843 NM_001031677.2 RAB24 ILMN_1773561 NM_021183.3 RAP2C ILMN_1749006 NM_052862.2 RCSD1 ILMN_2373266 NM_139168.2 SFRS12 ILMN_2117716 NM_005088.2 SFRS17A ILMN_2379835 NM_003352.4 SUMO1 ILMN_1712075 NM_015286.5 SYNM ILMN_1674866 NM_006354.2 TADA3L ILMN_2390227 NM_015043.3 TBC1D9B ILMN_1657983 NM_018975.2 TERF2IP ILMN_1705213 NM_022152.4 TMBIM1 ILMN_1756696 NM_207291.1 USF2 ILMN_2054442 NM_001099639.1 ZNF146 ILMN_2352590 NM_006974.2 ZNF33A

Subsequently, to confirm the effectiveness of self-normalization with reference genes, the correlation between quantile normalization for WG-DASL data and self-normalization data was investigated. Figure 1 shows the hazard ratios based on the two normalization methods. As a result, a strong correlation between quantile normalization and self-normalization methods was identified (Fig. 1).

Example 7: Development and evaluation of a prognostic prediction model based on gastric cancer prognostic genes-(1)

7-1: Prognosis Prediction Model Using Supervised Principal Component Analysis

To build a prognostic prediction model, a modified principal component analysis (SuperPC) developed by Bair and Tibshirani (PLoS Biol. 2004 Apr; 2(4): E108. Epub2004 Apr 13) was used. To develop and evaluate a gastric cancer prognosis prediction model based on SuperPC analysis, the BRB matrix tool (SimonR et al., Cancer Inform 2007; 3: 11-7) program developed by Richard Simon was used.

In SuperPC analysis, a p-value threshold for predicting prognosis at a desired level can be determined, and in the BRB matrix tools program, the default p-value is 0.001. The critical P-value can be less than 0.01 in any region, and with predefined p-values and calculated active components, the SuperPC analysis can include a subset of the prognostic genes listed in Table 2 and Table 3. To build a prognostic prediction model with proven validity, 10-fold cross-validation and SuperPC analysis were combined with the BRB matrix tool. As an example of SuperPC analysis, in order to establish a prognosis prediction model, a critical p-value of 0.00001 and two active components are used, and the SuperPC prognosis prediction model is composed of 7 prognosis genes, and this prediction model is shown in Figure 16 (Table 5 and Figure 16). In addition, Kaplan-Meier curves representing survival rates according to the expression levels of the 7 selected prognostic genes are shown in FIGS. 2 to 5 .

table 5

gene number gene symbol weight (wi) 1 ILMN_1713561 C20orf103 0.152677 2 ILMN_1672776 COL10A1 0.038261 3 ILMN_1663171 MATN3 0.016428 4 ILMN_1732158 FMO2 0.08681 5 ILMN_1811790 FOXS1 0.068965 6 ILMN_2402392 COL8A1 0.060799 7 ILMN_1736078 THBS4 0.088377

Figures 2 to 5 determine that, according to the expression levels of each of the seven genes listed in Table 5, the patient cohort was divided into a positive or negative prognosis group, and the survival rate of the positive prognosis group was compared with that of the negative prognosis group. The survival rate of the prognostic group appeared to be high. These results represent clinically that the prognosis of gastric cancer patients can be accurately predicted by measuring the expression levels of the gastric cancer prognostic genes in the present invention.

In addition, according to FIG. 16 , the results of establishing the prognosis prediction model of the 7 genes listed in Table 5 and dividing the patients according to the model showed that compared with the survival rate of the negative prognosis group (high risk), the survival rate of the patients divided into positive The group in the prognostic group (low risk) had a significantly higher survival rate, which corresponds to the actual clinical outcome (Figure 16). These results show that the 7 prognostic genes listed in Table 5 can be used to predict gastric cancer prognosis.

In addition, among the patients who have been classified according to the prognosis prediction model, the Ib/II stage gastric cancer patients are subdivided into a positive prognosis group or a negative prognosis group, and the Kaplan-Meier ratio representing the disease-free survival rate of the divided groups is shown in FIG. 17 . curve. As a result, the survival rate of stage Ib/II gastric cancer patients classified into the positive prognosis group according to the SuperPC prognostic prediction model was significantly higher compared to the survival rate of stage Ib/II gastric cancer patients classified into the negative prognosis group (Fig. 17).

Specifically, in the SuperPC prognostic prediction model (using the expression levels of 7 genes in Table 5), the prognostic index can be calculated by the following formula. If the determined prognosis index calculated by the following formula is greater than -0.077491, the patient from which the sample was taken can be classified in the negative prognosis group.

∑Iwi xi-4.51425

[wi and xi respectively represent the i-th weight (weight) and logarithmic expression level of the gene]

7-2: Comparative evaluation using conventional prognostic coefficients

The inventors used multivariate Cox analysis as a standard statistical analysis to determine whether prognostic predictions based on the prognostic genes of the present invention provide more meaningful prognostic information than conventional prognostic coefficients. Specifically, multivariate Cox models were investigated showing disease-free survival as assessed by the SuperPC prognostic index (Table 5) and 10-fold cross-validation, depth of tumor cell invasion (pT stage (pTstage)), tumor The number of lymph nodes to which cells metastasized (P Node).

Multivariate analysis identified seven prognostic genes independent of pT stage and P Node as excellent predictors of disease-free survival in gastric cancer patients undergoing curative gastrectomy and adjuvant chemotherapy (HR=1.9232, 95%CI, 1.4066 , 2.6294, P<0.0001, Table 6).

Table 6

covariate b SE P expect (b) 95% confidence interval for expectation (b) HER2 = Positive -0.3020 0.2510 0.2289 0.7393 0.4531 to 1.2062 PNODE 0.5548 0.08754 <0.0001 1.7415 1.4683 to 2.0657 Tstage 0.7339 0.1696 <0.0001 2.0831 1.4965 to 2.8997 Predicted Risk = "High" 0.6540 0.1604 <0.0001 1.9232 1.4066to2.6294

Example 8: Development and evaluation of a prognostic prediction model based on gastric cancer prognostic genes-(2)

8-1: Prognosis prediction model using gradient lasso method

The gradient lasso algorithm (Sohn I et al: Bioinformatics 2009; 25: 1775-81) was used to screen the 369 gastric cancer prognostic genes identified in Example 5 that could be used to predict the prognosis of gastric cancer. In the gradient lasso prognostic model, the prognostic score can be calculated using the following formula, and if the prognostic score of a random sample is positive, it can predict a positive prognosis.

is the regression coefficient estimated from the training set, and X is a vector of gene expression levels in the training set. ]

After gene selection using gradient lasso, susceptibility must be validated using an independent dataset. For this purpose, leave one out cross validation (LOOCV) is used. Specifically, leave-one-out cross-validation will use N-1 samples (training data), excluding one sample from the patient group (test data), when generating a prognostic prediction algorithm by gradient lasso, and will use The same sample was applied to the prognostic algorithm to classify the remaining one sample into the positive prognosis group or the negative prognosis group. This process is repeated for N samples of the patient group. After completing the division of all samples into the positive prognosis group or the negative prognosis group, the survival rate between the positive prognosis group and the negative prognosis group was compared by statistical analysis.

Twenty-six prognostic genes were screened by the gradient lasso algorithm during leave-one-out cross-validation and the screened genes are listed in Table 7. In addition, Kaplan-Meier curves representing the survival rate according to the expression levels of the 26 screened prognostic genes are shown in FIGS. 5 to 15 . According to Figures 5 to 15, it was determined that the patient group was divided into a positive prognosis group or a negative prognosis group according to the expression levels of each of the 26 genes listed in Table 7, and compared with the survival rate of the negative prognosis group, Survival rates appeared to be higher in the positive prognosis group. These results clinically represent that the prognosis of gastric cancer patients can be accurately predicted by measuring the expression levels of the gastric cancer prognostic genes in the present invention.

Table 7

Subsequently, the patient groups were divided into positive or negative prognosis groups according to this prognostic prediction model using 26 selected genes (gradient lasso and leave-one-out cross-validation). In addition, according to the pathological stage, the patient group that has been classified into the positive prognosis group or the negative prognosis group is subdivided, so that it is possible to predict the prognosis according to the pathological stage.

8-2: Evaluation of Prognostic Prediction Models Using the Gradient Lasso Method

To determine whether the prognosis predicted using the 26 prognostic genes coincided with the actual clinical outcome, the disease-free survival rates for the groups divided into positive and negative prognosis groups were shown in Kaplan-Meier curves (Figure 18). As a result, the 5-year disease-free survival rate was significantly higher in the positive prognosis group (low risk) compared to the 5-year disease-free survival rate in the negative prognosis group (high risk) (71.7% vs. Hazard ratio corresponding to a recurrence rate of 2.12 (95% CI, 1.57, 2.88, P=0.04, Figure 18). Therefore, it was determined that the prognosis of gastric cancer patients classified using 26 prognostic genes coincided with actual clinical outcomes.

To determine whether the prognostic results predicted by subdividing patients who had been divided into positive and negative prognostic groups according to pathological stage so that the prognosis could be predicted according to pathological stage coincided with actual clinical results, in Fig. 19 shows Kaplan-Meier curves representing disease-free survival by prognosis for groups of patients at each pathological stage. As a result, a cohort consisting of a total of 432 patients was divided into 145 patients in low-risk, stage Ib/II (5-year disease-free survival rate 84.8%); 90 patients in high Risk stage Ib/II (high-risk, stage Ib/II) (5-year disease-free survival rate was 61.1%); 83 patients were in low-risk stage III/IV (low-risk, stage III/IV) (5-year Disease-free survival rate was 48.9%), and 114 patients were in high-risk stage III/IV (high-risk, stage III/IV) (5-year disease-free survival rate was 36.9%). Specifically, it was determined that in stage Ib/II, the survival rate of the positive prognosis group (low risk Ib/II) was significantly higher compared to the survival rate of the negative prognosis group (high risk Ib/II), and even in stage III /Stage IV, the survival rate of the positive prognosis group (low risk III/IV) was significantly higher compared to the survival rate of the negative prognosis group (high risk III/IV) (Fig. 19).

The results showed that it is possible to accurately classify patients in pathological stages according to the prognosis by processing the expression levels of prognostic genes using statistical analysis algorithms, and the survival rate of gastric cancer patients can be improved by selecting appropriate treatment according to the predicted prognosis. For example, the expression levels of prognostic genes measured from patients diagnosed as stage Ib/II, self-normalized by measuring the relative expression levels relative to a reference gene, and, subsequently, if classified into the negative prognosis group Ib/II according to the gradient lasso algorithm If it is stage II, it can be determined that the prognosis of the patient is similar to that of stage III, and the survival of the patient can be prolonged by using the treatment for stage III patients.

8-3: Comparative evaluation using conventional prognostic coefficients

Known prognostic factors to predict the prognosis of gastric cancer are the determination of the depth of tumor cell invasion (pT stage) and the number of lymph nodes metastasized by tumor cells (P Node). The inventors used multivariate Cox analysis as a standard statistical analysis to determine whether prognostic predictions based on the prognostic genes of the present invention provide more meaningful prognostic information than conventional prognostic coefficients. Specifically, a multivariate Cox model was investigated showing disease-free survival, tumor cell invasion depth ( pT stage), the number of lymph nodes metastasized by tumor cells (P Node) or pathological stage (AJCC 6th edition). In this case, the pT period was divided into pT1/T2 and T3, and the logarithm of PNode was found by substituting 0.1 for 0.

Multivariate analysis identified 26 prognostic genes independent of pT stage and P Node as excellent predictors of disease-free survival in gastric cancer patients undergoing curative gastrectomy and adjuvant chemotherapy (HR = 1.859, 95% CI, 1.367 , 2.530, P=0.000078, Table 8). Likewise, as shown in Table 9, it was determined that 26 prognostic genes could independently predict disease-free survival at the last pathological stage (HR=1.773, 95% CI, 1.303, 2.413, P<0.00001, Table 9 The Pstage in is the combination of pTstage and PNode).

Table 8

Table 9

Example 9: Development and evaluation of a prognostic prediction model based on gastric cancer prognostic genes - (3)

9-1: Development and Evaluation of a Gastric Cancer Prognostic Score for Stage II Gastric Cancer Patients Using the nCounter Assay

By applying the gradient lasso algorithm to tumor samples from stage II gastric cancer patients (N=186) obtained from the cohort used within the WG-DASL assay, 8 gastric cancer prognostic genes providing robust prognostic information were identified. combination (Table 10). The Gastric Cancer Prognostic Score (GCPS) was developed using a linear combination of the normalized expression levels of the 8 genes and the Cox regression estimates. Measurement of gene expression levels was performed using the nCounter Assay Kit (system; NanoString Technologies).

Table 10

gene symbol regression estimate C20orf103 0.0636 CDC25B -0.0175 CDK1 -0.1005 CLIP4 0.4822 LTB4R2 -03950 MATN3 0.2982 NOX4 0.0288 TFDP1 -0.2886

The GCPS assigning 25% of patients to the negative prognosis group was determined to be the most robust by analyzing the cut-off (Figure 28). This cutoff was chosen for validation in a future independent validation cohort. As a result of applying optimized cut-offs to this cohort, as shown in Figure 29, compared to the 5-year disease-free survival rate of 84.3% for the low-risk group (top graph), based on the predictive model, high The 5-year disease-free survival for the risk group (bottom panel) was 42.6% (p<0.0001).

Due to the overfitting problem, it was necessary to validate the GCPS with a revised algorithm and cutoffs using an independent patient cohort that was not used to identify genes. For this purpose, a patient cohort for validation is first obtained. GPS was applied to an independent validation cohort of stage 2 gastric cancer patients who received the same chemo-radiotherapy as those used to identify patients with gastric cancer prognostic genes (n=186, discovery cohort). As a result, the risk score distribution is very similar to Figure 30, which represents the robust analytical performance of this assay.

Applying the predefined GCPS cut-off value (0.2205) obtained from the discovery cohort to the validation cohort and generating Kaplan-Meier curves based on rank distributions, this algorithm can accurately identify gastric cancer risk in stage 2 gastric cancer patients receiving chemoradiation. High patients (Figure 31). As shown in Figure 31, GPS of these eight prognostic genes successfully predicted 216 patients in the high-risk group (5-year DFS, 58.7%, bottom graph) and low-risk group (5-year DFS, 86.3%, top graph). Patients with stage 2 gastric cancer (P=.00004, HR=3.15).

9-2: Optimization of GCPS

According to this example, after verifying that high-risk patients among stage 2 patients receiving chemoradiotherapy could be identified by the expression profile of gastric cancer prognostic genes, the discovery cohort and validation cohort were merged into one cohort to develop the second-generation GCPS. To develop a predictive model based on the Cox proportional hazards model of disease-free survival, the gradient lasso (least absolute shrinkage and selection operator) algorithm was used. Table 11 represents the 13 genes (probes) constituting the prediction model obtained when using a phase 2 data set (phase2 data set, N=402), wherein the phase 2 data set is a combination of a discovery set and a validation set.

Table 11

gene symbol regression estimate ADRA2C -0.0156 C20orf103 0.1082 CLIP4 0.3891 CSK -0.6654 FZD9 -0.0829 GALR1 -0.0509 GRM6 -0.0244 INSR 0.0251 LPHN1 -0.0126 LYN -0.0012 MATN3 0.2134 MRGPRX3 -0.0009 NOX4 0.0951

The patient's GCPS was calculated as [S=β ₁ x ₁ +...+β _n x _n ]. Wherein, x _n is the quantitative expression value of the nth gene, β _n is the regression estimate (Regression estimate) of the nth gene listed in Table 10 and Table 11, and S represents the gastric cancer prognosis score. Subsequently, critical values for the first and third quartiles of the risk score distribution (Q1 = -0.9842, Q3 = -0.4478) were estimated from the Phase 2 dataset. By applying this cutoff to the final validation set of 306 patients, patients with GCPS below Q1 and patients with GCPS above Q3 were assigned to low-risk and high-risk groups, respectively. As a result, as shown in FIG. 24 , the Kaplan-Meier curve determined that the survival rate (bottom graph) of patients in the group predicted to be high risk was significantly lower compared to patients in other groups ( FIG. 32 ).

9-3: Validation of second-generation GCPS in surgery-only stage II gastric cancer patients

To test the performance of GCPS on patients who underwent only surgery without chemotherapy or radiation, cancer tissue from 306 patients diagnosed with stage 2 gastric cancer who underwent only radical gastrectomy at Samsung Medical Center was examined Without receiving adjuvant chemotherapy (adjuvant chemotherapy) or radiation therapy after surgery. Patients were selected according to the following criteria.

Among 476 gastric cancer patients diagnosed with stage 2 pathological stage who underwent curative gastrectomy alone without adjuvant chemotherapy or surgery at Samsung Medical Center from April 1995 to September 2006 After radiotherapy), 306 patients were selected according to the following criteria.

1) Histological diagnosis of adenoma,

2) tumor resection, no residual tumor,

3) D2 lymph node dissection,

4) At least 18 years old,

5) According to AJCC (American Joint Committee on Cancer) 6th edition, pathological stage II (T1N2, T2aN1, T2bN1 and T3N0),

6) Completely keep the operation records and treatment records.

170 patients in a cohort of 476 patients were excluded from this analysis due to the following reasons:

1) Patients without complete medical records (N=66),

2) Disease-free death or unexplained death (N=43),

3) Corrected pathological staging (N=45)

4) Paraffin blocks are not available (N=15),

5) Double primary cancer (N=1).

As shown in Figure 33, as a result of application of second-generation GCPS to a cohort of patients with stage 2 gastric cancer who underwent surgery only, although GCPS was developed using a cohort of patients who Patients in the high-risk group (bottom panel) were identified as having poor prognosis (p=0.00287). This result suggests that high-risk patients defined by GCPS essentially have a poor prognosis that is not improved by anticancer drugs and radiotherapy, and that new therapies need to be developed for these patients.

Claims (24)

1. A marker for predicting the prognosis of gastric cancer, comprising one or more genes selected from the following genes: C20orf103 (chromosome 20 open reading frame 103), COL10A1 (type X collagen α1), MATN3 (maternal protein 3) , FMO2 (flavin-containing monooxygenase 2), FOXS1 (forkhead box protein S1), COL8A1 (type VIII collagen α1), THBS4 (platelet response protein 4), CDC25B (cytokinesin 25 homologue B ), CDK1 (cyclin-dependent kinase 1), CLIP4 (adapter protein family member 4 containing CAP-GLY domain), LTB4R2 (leukotriene B4 receptor 2), NOX4 (NADPH oxidase 4), TFDP1 ( Transcription factor Dp-1), ADRA2C (α-2C-adrenergic receptor), CSK (c-src tyrosine kinase), FZD9 (Frizzled family receptor 9), GALR1 (galanin receptor 1), GRM6 (Glutamate receptor metabotropic type 6), INSR (Insulin receptor), LPHN1 (Spider toxin receptor 1), LYN (v-yes-1 Yamaguchi sarcoma virus-related oncogene homolog), MRGPRX3 (MAS Related GPR member X3), ALAS1 (delta-aminolevulinic acid synthase 1), CASP8 (caspase 8, apoptosis-associated cysteine peptidase), CLYBL (citrate lyase-like beta), CST2 (cysteine protease inhibitor SA), HSPC159 (galectin-binding lectin), MADCAM1 (mucosal angioaddressin cell adhesion molecule 1), MAF (v-maf aponeurotic fibrosarcoma oncogene homolog ( avian), REG3A (regenerating islet-derived protein 3α), RNF152 (ring finger protein 152), UCHL1 (ubiquitin carboxy-terminal esterase L1 (ubiquitin thioesterase)), ZBED5 (zinc finger containing BED-type protein 5), GPNMB (glycoprotein (transmembrane) nmb), HIST1H2AJ (histone cluster 1, H2aj), RPL9 (ribosomal protein L9), DPP6 (dipeptidyl peptidase 6), ARL10 (ADP-ribosylated Factor-like 10), ISLR2 (leucine-rich repeat immunoglobulin superfamily 2), GPBAR1 (G protein-coupled bile acid receptor 1), CPS1 (carbamoyl-phosphate synthase 1), BCL11B (B -CLL/Lymphoma 11B (zinc finger protein)) and PCDHGA8 (protocadherin gamma subfamily A8) genes. 2. The marker according to claim 1, wherein said marker is the C20orf103, COL10A1, MATN3, FMO2, FOXS1, COL8A1 and THBS4 genes. 3. The marker according to claim 1, wherein the marker is ALAS1, C20orf103, CASP8, CLYBL, COL10A1, CST2, FMO2, FOXS1, HSPC159, MADCAM1, MAF, REG3A, RNF152, THBS4, UCHL1, ZBED5, GPNMB, HIST1H2AJ, RPL9, DPP6, ARL10, ISLR2, GPBAR1, CPS1, BCL11B, and PCDHGA8 genes. 4. The marker of claim 1, wherein the marker is the C20orf103, CDC25B, CDK1, CLIP4, LTB4R2, MATN3, NOX4, and TFDP1 genes. 5. The marker of claim 1, wherein the marker is ADRA2C, C20orf103, CLIP4, CSK, FZD9, GALR1, GRM6, INSR, LPHN1, LYN, MATN3, MRGPRX3 and NOX4 genes. 6. The marker of claim 1, wherein the gastric cancer is stage Ib or II gastric cancer. 7. A composition for predicting the prognosis of gastric cancer, comprising a reagent for measuring the expression level of mRNA or protein of the marker for predicting the prognosis of gastric cancer according to any one of claims 1 to 6. 8. The composition according to claim 7, wherein the reagent for measuring the mRNA expression level of the gene is a primer pair or a probe specifically binding to the gene. 9. The composition of claim 7, wherein the reagent for measuring the expression level of the protein is an antibody specific for the protein encoded by the gene. 10. A kit for predicting the prognosis of gastric cancer, comprising a reagent for measuring the expression level of the mRNA or protein of the marker for predicting the prognosis of gastric cancer according to any one of claims 1 to 6. 11. A method for predicting the prognosis of gastric cancer, comprising a) obtaining the expression level of the mRNA or protein of the marker for predicting the prognosis of gastric cancer according to any one of claims 1 to 6 in a sample collected from a gastric cancer patient or expression pattern; and b) comparing the expression level or expression pattern obtained in step a) with the expression level or expression pattern of mRNA or protein of the corresponding gene in gastric cancer patients whose prognosis is known. 12. The method of claim 11, wherein the sample is gastric tumor tissue. 13. The method according to claim 11, wherein the expression level of the mRNA of the gene is measured by using a primer pair or a probe specifically combined with the gene and the expression level of the protein is measured by using an antibody specific to the corresponding protein. The expression level. 14. The method according to claim 11 , wherein the mRNA expression level of one or more reference genes selected from the reference genes listed in Table 4 or the expression level of the protein encoded by the reference gene For comparison, the expression levels of the mRNA of the gene or the protein encoded by the gene are normalized. 15. The method according to claim 11, wherein the gastric cancer patient in steps a) and b) is a patient receiving the same treatment, and said treatment is selected from radiotherapy, chemotherapy, chemoradiotherapy, adjuvant chemotherapy, gastrectomy , chemotherapy or chemoradiation after gastrectomy, and gastrectomy without radiation therapy after adjuvant chemotherapy or surgery. 16. A method for predicting the prognosis of gastric cancer, comprising a) measuring the mRNA or protein expression level of the marker for predicting the prognosis of gastric cancer according to any one of claims 1 to 6 in a sample collected from a gastric cancer patient or b) apply the expression value obtained in step a) to the prognosis prediction model to obtain the gastric cancer prognosis score; and c) compare the gastric cancer prognosis score obtained in step b) with the reference value to obtain The prognosis of the patient is determined. 17. The method according to claim 16, wherein in step a) C20orf103, CDC25B, CDK1, CLIP4, LTB4R2, MATN3, NOX4 and TFDP1 genes; or ADRA2C, C20orf103, CLIP4, CSK, FZD9, GALR1, GRM6, Expression levels of mRNA or protein of INSR, LPHN1, LYN, MATN3, MRGPRX3 and NOX4 genes. 18. The method according to claim 16, wherein the prognosis prediction model in step b) is expressed as: [S=β ₁ x ₁ +...+β _n x _n ] Among them, x _n is the quantitative expression value of the nth gene, _βn is the Cox regression estimate for the nth gene, and S stands for gastric cancer prognosis score. 19. The method according to claim 16, wherein the reference value in step c) is defined as the third quartile critical value to the fourth quartile critical value in the distribution of gastric cancer prognosis score Value, the gastric cancer prognosis score is obtained from multiple gastric cancer patients according to the following formula: [S=β ₁ x ₁ +...+β _n x _n ] Among them, x _n is the quantitative expression value of the nth gene, _βn is the Cox regression estimate for the nth gene, and S stands for gastric cancer prognosis score. 20. The method according to claim 16, wherein the reference value in step c) is defined as the value within the range of the second quartile critical value to the third quartile critical value in the gastric cancer prognostic score distribution Value, the gastric cancer prognosis score is obtained from multiple gastric cancer patients according to the following formula: [S=β ₁ x ₁ +...+β _n x _n ] Among them, x _n is the quantitative expression value of the nth gene, _βn is the Cox regression estimate for the nth gene, and S stands for gastric cancer prognosis score. 21. The method according to claim 16, wherein it is determined that a case having a gastric cancer prognostic score obtained in step b) which is equal to or greater than the reference value has a negative prognosis. 22. The method according to claim 19, wherein the cut-off value of the third quartile is 0.2205 or -0.4478, and it is determined to have a gastric cancer equal to or greater than the cut-off value obtained in step b) Cases with a prognostic score had a negative prognosis. 23. The method according to claim 22, wherein in the case of measuring the expression levels of C20orf103, CDC25B, CDK1, CLIP4, LTB4R2, MATN3, NOX4 and TFDP1 genes in step a), the cutoff value is 0.2205. 24. The method according to claim 22, wherein in the case of measuring the expression levels of ADRA2C, C20orf103, CLIP4, CSK, FZD9, GALR1, GRM6, INSR, LPHN1, LYN, MATN3, MRGPRX3 and NOX4 genes in step a) , the critical value is -0.4478.

Images